Skip to main content

Full text of "The uniqueness of the individual"

See other formats

|i: ■:f 










The Uniqueness 
of the Individual 



Jodrell Professor of Zoology and Comparative Anatomy 
University College, University of London 


First printing 1957 
Second printing 1958 

Published in Great Britain by Methuen & Co. Ltd. 

Library of Congress Catalog Card No. 57-10538 

Manufactured in the United States of America 


Introduction P^g^ ^^ 

1. Old Age and Natural Death (1946) 17 

2. An Unsolved Problem of Biology (1952) 44 

3. A Note on ^The Scientific Method^ (1949) 71 

4. A Commentary on Lamarckism (1953) 79 

5. The Pattern of Organic Growth and Trans- 
formation (1954) 108 

6. 'The Imperfections of Man (1955) 122 

7. Tradition: The Evidence of Biology (1953) 134 

8. The Uniqueness of the Individual (1956) 143 
Index 186 



I thank their respective pubhshers for permission to reproduce 
the following articles: Old Age and Natural Death: Modern 
Quarterly^ vol. 1 (1946), p. 30. An Unsolved Problem of Biology: 
H. K. Lewis, London, 1952. A Note on 'The Scientific Method^ 
Nineteenth Century (1949), p. 115. A Commentary on Lamarck- 
ism: Bull. Nat. Inst. Sci. India, vol. 7 (1953), p. 127. The 
Pattern of Organic Growth and Transformation: The Times 
Science Review.^ no. 12 (1954), p. 10. The Imperfections of 
Man: Twentieth Century (1955), p. 236. 


There are collected in this volume some of the articles or 
lectures I have written or delivered over the past ten years. 
Most of them deal with problems of evolution, a choice which 
may be thought to need a word of justification. Not so long ago 
there was a slump in research and thought on evolution. 
Darwin''s successors had made it their business to reveal and 
expound the detailed progress of evolution, but did not feel 
obliged to commit themselves to any particular theory of its 
mechanism. Their work reached its peak with the publication 
of the majestic but unfinished Treatise of Zoology^ under Ray 
Lankester's editorship, shortly before the First World War; it 
was founded squarely upon the concept of '"homology"', i.e. of 
the evolutionary connections between parts of animals — 
between fins and wings and limbs, for example — rather than 
between animals considered as a whole. But between that 
Augustan age of comparative anatomy and the rethinking of 
Darwinism in the language of genetics, almost no progress was 
made in the understanding of the evolutionary mechanism. 
Many biologists became querulous and uneasy about the pre- 
vailing Darwinian theory — a dissatisfaction nowhere to be 
more clearly seen than in that great baroque masterpiece of 
biological literature, D''Arcy Thompson'^s essay On Growth and 
Form (1915). Laymen were therefore to be forgiven if they 
thought that Darwinism had been discredited or had died of 
inanition. The pity is that, in spite of the advocacy of two 
generations of Huxleys, many educated laymen hold that 
opinion still, although any good ground for doing so has long 
since disappeared. 



Research on evolution was never so lively and pointed as it 
is to-day. It runs along several different lines. First, beginning 
in the 1920s, was the refounding of Darwinism upon the prin- 
ciples of Mendelian genetics — the formulation of a genetical 
theory of natural selection — and this was mainly the work 
of R. A. Fisher, J. B. S. Haldane and Sewall Wright, though 
the names of Norton and Lotka are by no means to be for- 
gotten. At first these endeavours were purely theoretical, but 
in due course, and with all the economy of labour that only a 
deep theoretical grasp of the problem could have made possible, 
its predictions were tested in the laboratory and in the field. 
(It is indeed only in genetics that there can be said to be a 
""theoretical biology**, in the sense in which theoretical physics 
is so described.) Thanks very largely to the work of the new 
Oxford school of evolutionists under E. B. Ford it is now 
possible to witness the act and not merely the accomplished 
fact of evolution; we have now an altogether new understand- 
ing of the extraordinary delicacy and responsiveness with 
which the genetical structure of a population can be remoulded 
by selective forces. Standing a little aside from these accom- 
plishments, because it took origin from the study of develop- 
ment rather than of heredity, is C. H. Waddington''s important 
demonstration of how a change which was at first brought about 
by the action of the environment may, under the influence of 
selection, become genetically ingrained: habit becoming herit- 
age, or nurture nature, as you will. Then, finally, there are the 
grand speculations on evolution that perpetuate an older 
tradition of thought. For example, it has become ever clearer, 
since Garstang and de Beer first turned our thoughts in that 
direction, that paedomorphosis is a fundamental stratagem of 
evolution — that animals can, so to speak, slough away the 
latter ends of their life histories and build their lineages anew 
upon larval or even embryonic forms. There can be little reason- 
able doubt that vertebrate animals arose in this manner from 



ancestors akin to sea-squirts, and man may have done so too, 
though less dramatically, from ancestors akin to apes; for the 
fundamental pattern of a vertebrate is indeed that of the larva 
of a sea-squirt, upon which the feeding mechanism of the adult 
has somehow been superimposed; and the story of how man 
may have taken origin from a foetal ape became science fiction 
not so long after it made claim to being science fact. But I feel 
obliged to report that speculation on paedomorphosis is tend- 
ing to get a little out of hand. In the old days, if animal A had 
a B-Vike stage in its life history, that was taken as certain 
evidence that A had evolved from B and that its transiently 
B-Yike condition was an example of HaeckePs '"Law of Re- 
capitulation'', viz. that an animal, as it develops, climbs up its 
own family tree. To-day no one believes in recapitulation in 
this simple form, and when A has a j5-like stage in its life 
history, it tempts people to declare that B arose bv paedo- 
morphosis from an ancestor akin to A. All the evidence that 
was at one time thought to exemplify recapitulation is now 
inverted and used as e\'idence of a paedomorphic origin: 
Haeckel is still the hero, though his portrait hangs upside 
down: but neither way up can Haeckel wield his old authority 
without better evidence than his successors have so far been 
able to provide. 

Nothing in evolutionary thought shows up the difference 
between professional and layman so clearly as their attitude 
towards Lamarckism, the subject of the fourth of the essavs 
printed here. Nearly all laymen, and most young zoologists 
before they learn better, believe implicitly in what has come to 
be called, no matter how inappropriately, the ''inheritance of 
acquired characters\ It is an intelligible and forthright doc- 
trine, and, in ignorance of genetics, an alternative is difficult 
to propound; more than that, a certain deep-seated sense of the 
fitness of things is gratified by the belief than an animaPs own 
activities, accomplishments and endeavours should contribute 



to the heritage of its young. Lamarckism is a theory that 
deserves to be taken seriously, and in my Commentary on the 
subject I have striven to do so. But it is not true; the genetical 
system of an individual cannot take the impress of the environ- 
ment; it is a population, not a pedigree, that evolves. Modern 
Darwinism demands that the genetical system of an individual 
should be highly stable, as indeed it is. The stability is not 
absolute, it is true, but departures from it are of a special kind 
— rare, abrupt and discontinuous; they do not arise in response 
to an organism''s needs, nor do they, except by accident, gratify 
them. And even the ''instability' of the genetic mechanism is 
best thought of as a sudden passage from one stable state to 
another, for a mutant gene, once it has arisen, perpetuates 
itself faithfully in its altered form. It is deeply necessary, for 
any clear understanding of evolution, to distinguish between 
the genetical structure of a population, which is quite wonder- 
fully malleable and responsive to the impress of the environ- 
ment, and the genetical make-up of an individual, which, in a 
physico-chemical sense, as Schrodinger has told us, is almost 
miraculously stable. 

One of the most popular misconceptions about the theory of 
evolution by natural selection is that which treats it as the 
denouement of the following train of thought: (a) organisms 
produce offspring in numbers vastly in excess of their needs; 
{h) only a minority survive; therefore (c) only those survive 
which are best equipped to do so, the ''fittest\ The catch in 
this Malthusian syllogism, pointed out years ago by Fisher, 
lies in its major premise {a). So far from producing a vastly 
excessive number of offspring, most organisms produce just 
about that number which is sufficient and necessary to per- 
petuate their kind. Degree of fecundity is one of the conse- 
quences of natural selection: it is not its cause. Nidicolous 
birds, Lack has shown us, illustrate this truth with particular 
clarity; they do in fact lay clutches of a certain size, though 



they could lay more eggs — the egg industry is founded upon the 
inexhaustible gullibility of the domestic fowl — and could, of 
course, lay fewer. Having regard to all the exigencies of giving 
birth to and rearing eggs and young, the size of its clutch is just 
about that which gives each species its greatest likelihood of 
self-perpetuation . 

A second misconception may be aptly called the Zenonian, 
because of a certain family likeness to the argument which 
purports to show that Achilles can never overtake the tortoise, 
nor an arrow reach its target. Any substantial adaptation, it 
is argued, can only be achieved by the adding up, over very 
many generations, of single all but infinitesimal adaptative 
changes which, being of inappreciable advantage to their 
owners, offer nothing for selection to get to grips upon. Luckily 
selection does not abide by human judgements of its efficacy; 
it can be shown that even so slight a selective advantage as that 
which allows one thousand and one of its possessors to per- 
petuate themselves for every thousand that lack it, must 
eventually prevail. 

Philosophers have sometimes beguiled themselves with a 
third argument allegedly discreditable to Darwinism. Selection 
(the argument runs) can select only from what is available and 
"'given'', i.e. can select only from within the compass of existing 
variation and diversity; by selection, all men or all deer could 
be caused to become as tall or swift as the tallest or swiftest 
now among them, but not still taller or swifter still. The error 
here is to equate all variation with overt variation, to suppose 
that all genetic goods are in the shop window. The genetical 
mechanism is such that there are deep resources of hidden 
variation, of possible animals only awaiting the occasion to 
become real. Even if mutation were to take a holiday, as Death 
did in Casella"'s play, evolution would certainly not come to a 
standstill; how far it could go we obviously cannot say. 

I think I have given reasons enough to justify a certain pre- 



occupation with problems of evolution. The first two essays 
have to do with the evolution of *" ageing', i.e. of senescence, and 
the second can be regarded as a lengthy footnote to the latter 
part of the first; but the reader should be warned that unlike 
some of my predecessors I have not thought it necessary to pull 
a long face or strike pious attitudes when writing of senescence 
and old age. A Commentary on Lamar ckism has already been 
referred to; The Imperfections of Man argues that evolution 
must always be a compromise in which a lesser evil is put up 
with to make possible a greater good. In Tradition: The Evidence 
of Biology I deal briefly with the idea that man enjoys a new 
modality of evolution, one in which tradition takes the place 
of heredity iYi conferring upon man the qualities that make 
him '■fitter'' than other animals and his hope of becoming fitter 
still. The Uniqueriess of the Individual is about the sources and 
nature of inborn diversity. Two other essays are added for full 
measure: A Note on ''The Scientific Method''^ which shows to 
what a large extent we are all indebted to the reasoning of 
Karl Popper, however imperfectly the debt is paid, and The 
Patterii of Organic Growth ajid Transformation — too long a 
title, perhaps, for an essay which could only be broad in com- 
pass by resisting the temptation to go deep. 

The last essay, on The Uniqueness of the Individual^ was 
written specially for this volume, and deals with the kind of 
research I do myself; the sources of the remainder are listed in 
the Table of Acknowledgements. 

Looking back on these essays, I see their imperfections much 
more clearly than their merits, but I have tried to repair some 
of the more obvious errors of judgement, or errors or omissions 
of fact, by additional footnotes, set in square brackets. 


' • ' MASS. 

Old Age and Natural Death 

The problems of old age and natural death are hardly yet 
acknowledged to be within the province of genuine scientific 
enquiry. This does not mean that biologists are ignorant of the 
fact that such problems exist, nor that natural death is 
altogether insusceptible of scientific treatment. It simply means 
that no such treatment has been given it yet.* 

This neglect is partly the outcome of a certain quickening 
in the tempo of biological research. The biologist of to-day is a 
busy man: he has no time for anecdotes about the age of 
tortoises, and wants more evidence than MetchnikofF had 
power to give him before he takes steps to modify the flora of 
his bowel. Yet nearly all the great theorists of the last century 
were fond of teasing themselves with speculations about death. 
''Qu'est ce que la vief Claude Bernard^ asked himself: ''La vie, 
c'est la mort.'' Life is combustion, and combustion death. ''La 
vie est un minotaure, elle devore Vorganisme.'' This is only one of 
alternative views on the nature of natural death. The distinc- 
tion of first suggesting that natural death might be an epi- 
phenomenon of life, rather than something of the very nature 

^ Definition de la Vie (1875); one of the essays reprinted in La Science 
Experimentale, 7th ed., Paris, 1925. Also Bernard quoting Buffon. 

* [This is no longer true; there has now been a notable revival of interest 
in senescence, supported in England by the Nuffield and Ciba Foundations, 
and owing much to the apostolic vigour of Dr V. Korenchevsky. Alex 
Comfort's The Biology of Senescence (London, 1956) summarizes much of 
the work that has been published since this article was written.] 

B 17 


of the act of living, is shared unequally between August Weis- 
mann^ and Alfred Russell Wallace. (Wallace''s views are known 
to us only from a casual letter preserved by Edward Poulton.^ 
They are about the same as Weismann''s, though less confid- 
ently and much less lengthily expressed.) But before I try to 
give an account of Weismann''s views, we must have a few 
definitions; for the trouble with ""natural death' is not that it 
lacks a meaning, but that it has the embarrassment of two or 
three. By ""accidental death*', then, or simply '"death"', is meant 
death from any cause whatsoever. *"Natural death** is that sort 
of death by accident to which the age-specific decline of our 
faculties, senescence^ has made a certain contribution, however 
small. The contribution grows larger as we grow older: what 
lays a young man up may lay his senior out: but it always falls 
short of unity, for no one dies merely of the weight of years. 
The greatest clinical pathologist of the last generation' looked 
back upon his life for evidence of such a case. He once thought 
he had found it in a colleague ninety-four years old, whose life 
seemed merely to fade away; but autopsy showed a lobar 
pneumonia of four days'* standing. 

We shall be obliged to use the term '"natural death"* in the 
rather wide sense of the foregoing definition. Popular usage 
quite rightly fines it down to forms of death to which senes- 
cence has made a pretty big contribution, for it seems absurd 
to say that a man of forty could die in part of old age. At all 
events, what we are to discuss is not the event, death, but the 
process of senescence. (The definition of death itself, in the 
most familiar of its several meanings, can be valid only with 
reference to some stated ""leveP of biological organization. A 
society will die before its individual members, an individual 

^ The Duration of Life (pp. 1-66) and lAfe and Death (pp. 111-61); essays 
reprinted in Weismann on Heredity^ ed. E. B. Poulton, S. Schonland, and 
A. E. Shipley; 2nd ed., Oxford, 1891. 

2 The Duration of Life, pp. 23-4. » Cf. Lancet, 235, p. 87, 1938. 



before his cells, and a cell before its ferments have stopped 
working. But legally, I suppose, a man is dead when he has 
undergone irreversible changes of a type that make it im- 
possible for him to seek to litigate.) 

Weismann believed that natural death had evolved under a 
Darwinian regimen of natural selection. The '"utility of death"* 
he says, is this. ""Death takes place because a worn-out tissue 
cannot for ever renew itself. . . . Worn-out individuals are not 
only valueless to the species, but they are even harmful, for 
they take the place of those which are sound.' It follows that 
'by the operation of natural selection, the life of a theoretically 
immortal individual would be shortened by the amount which 
was useless to the species\^ In this short passage, Weismann 
canters twice round the perimeter of a vicious circle. By 
assuming that the elders of his race are decrepit and worn out, 
he assumes all but a fraction of what he has set himself to 
prove. Nor can these dotard animals ""take the place of those 
which are sound** if natural selection is working, as he tells us, 
in just the opposite sense. It is curious that Metalnikov in his 
comparatively recent La Lutte contre la Mort (1937) should 
give these fallacies a seventy-five-year run by twice repeating 
them with approval word for word. The problem is, why are the 
older animals decrepit and worn out? And for this Weismann 
had no sufficient answer. It must be obvious that, senescence 
apart, old animals have the advantage of young. For one thing, 
they are wiser. The Eldest Oyster, we remember, lived where 
his juniors perished. They are wiser, too, in their experience of 
infection, for an animal which has survived a first infection is 
better equipped to deal with it a second time. In the majority 
of animals 'immunological wisdom** may be a better bargain 
than anything they may have by way of mind. We are always 
inclined to over-estimate the value of mental wisdom, though 
no one, I suppose, has the temerity to doubt that the giraffe 

^ The Duration of Life ^ p. 24. 



owes more to his long neck than to the organ poised on top of 
it; and the logic of brute fact tells us that the extinct reptile, 
Diplodocus^ which had a brain in the pelvic region as well as up 
in front, drew little advantage from his power to reason not 
merely a priori. No: what kills the old animal is not in the first 
place decrepitude, but something which has the dimensions of 
the product of time by luck. 

Weismann had a theory not merely of the evolution of death 
in animal populations but also of the mechanism of senescence 
in the individual. He believed that a limit to life was set by an 
inherent limitation in the power of germ cells to divide. ""We do 
not know', he says, Svhy a cell must divide 10,000 or 100,000 
times and then suddenly stop,''^ as he thought it did. As a 
matter of fact, we now know that no such inherent limitation 
exists; but Weismann'^s theory — if we disregard the fact that 
the progeny of a cell which divided only 10,000 times would fill 
the utter limits of known space — shows that he had not 
appreciated the ""asymptotic'' character of the process of age- 
decline. He had no grasp of the process of ageing. We don''t 
grow old suddenly, and the cells within us do not suddenly stop 
dividing. Those that do stop come to rest in a decent orderly 
fashion. Charles Minot^ was the first to make this clear. He 
took over Weismann''s idea that death had evolved by natural 
selection, and turned his mind to ageing in the individual 
alone. His views were original and still are theoretically import- 
ant, so they deserve a fuller treatment than they commonly 

Minot used growth as a measure of vitality; not the mere 
rate of growth, but the specific rate, which gives us a measure 
of the capacity of living tissue to reproduce itself at the rate 
at which it was formed. It is simply the rate of growth at any 

1 Ibid., p. 22. 

2 The Problem of Age, Growth, and Death, Ldndon, 1908; a series of 
lectures first published in Popular Science Monthly. 



chosen time divided by the size at that same time — in other 
words, by the material theoretically available for further grow- 
ing. It cannot be denied that the specific growth rate is a 
measure of vitality, though not perhaps so complete a measure 
as Minot in his time believed. Minot found that the power of 
tissue to reproduce itself at the rate at which it was formed fell 
off through life from earliest childhood onwards. He found that 
the decline was faster in children than in their elders, and, 
indeed, that it fell off more and more slowly as life went on. The 
inferences he drew were these. There is no period of increasing 
vitality leading to the mature state and thereafter to the senile; 
the process of ageing goes on continuously throughout life. 
And ageing is faster in young animals than their elders — ""a 
strange, paradoxical statement\ ^Our notion that man passes 
through a period of development and a period of decline is 
misleading ... in reality we begin with a period of extremely 
rapid decline, and then end life with a decline which is very 
slow and very slight."* 

This is a good moment to ask what the life insurance com- 
panies have to say about these problems. Their evidence is at 
first sight very helpful. Look at the curve from which the 
actuary computes the force of mortality at various ages — the 
curve which defines, for each age of life, the numbers still living 
of a certain initial number born alive. ^ From the twelfth or 
fifteenth year onwards in human life, the curve is smooth; there 
is no break or discontinuity, no hint at all that at such an age 
the prime of life has ended and old age begins. Nor is this 
generalization false for animals other than man. *'Life tables' 
for them are pitifully meagre; but Leslie and Ranson made one 

^ For the terminology used in actuarial work, cf. L. Hogben: The Measure- 
ment of Human Survival, in New Biology, ed. M. Abercrombie and M. L. 
Johnson. Penguin Books, London, 1944. [See also L. I. Dublin, A. J. 
Lotka and M. Spiegelman, Length of Life: a Study of the Life Table, New 
York, 1949.] 

/>w J. 


lately for the laboratory vole,^ and here too we find the same 
smooth passage to extinction. *" Voles drop off at all times of 
life,"* says Elton,^ speaking of this evidence, 

'though not at the same rates. And these are not "ecologicar* 
deaths; few of them probably are "parasitological" deaths. We 
hardly know what process is at work, and for want of a better term 
we may call it "wear and tear". This has the suggestion of an 
internal breakdown in the physiological organization. We might 
almost say that the process of senescence begins at birth.' 

This final inference, which I have italicized, is by no means 
immediate. The actuary's life table is not a mapping of the 
course of individual life: it is founded on the distribution 
through life of the ages at which people die. It thus relates to 
no event in life save one, its end. Even if the sudden flowering 
of an evil gene caused voles to age and die within a day, the 
ages of their deaths might well be so pieced out among the 
population as to yield just that smooth, continuous curve the 
actuary maps for us. If, however, the population is reasonably 
uniform, then the life table (or rather, the force of mortality 
computed from it) does indeed give us what may be called a 
•"statistical picture** of the course of ageing. For we may define 
'senescence"* as that which predisposes the individual to death 
from accidental causes of random incidence; and it follows that 
the frequency distribution of the ages of death gives us a 

^ P. H. Leshe and R. M. Ranson, Journal of Animal Ecology ^ 9, p. 27, 
1940. For life tables for invertebrate animals, cf. A. J. Lotka, The Elements 
of Physical Biology, Baltimore, 1925; W. H. Dowdeswell, R. A. Fisher, and 
E. B. Ford, Annals of Eugenics, 10, p. 123, 1940; C. H. N. Jackson, ibid., 
p. 832. Jackson finds that the life table of tsetse flies is biased, during the 
rainy season only, by an element contributed by senescence. [For further 
evidence see Principles of Animal Ecology, by W. C. Allee, O. Park, A. E. 
Emerson, T. Park and K. P. Schmidt (Philadelphia, 1949); E. S. Deevey, 
Quart. Rev. Biol., 22, p. 283, 1947; The Natural Regulation of Animal 
Numbers, by David Lack (Oxford, 1954); The Biology of Senescence, by 
Alex Comfort (London, 1956).] 

2 C. S. Elton, Voles, Mice and Lemmings, pp. 202-5, Oxford, 1942. 



statistical picture of the magnitude of this predisposition. 
Many sciences use a picture of this sort, and some use no other; 
the problems it raises are interesting, but not at the moment 

Minot wanted to bring not merely size but shape as well 
within the ambit of his laws; but complained, as many have 
done since, that ""we do not possess any method of measuring 
differentiation which enables us to state it numerically\ Such 
attempts as have been made to do so support his theory; for 
example, the rate of change of shape of the human being falls 
off progressively through life.^ But we do know that Minofs 
laws are by no means commonly true of faculties other than 
those which turn upon the pattern and the rate of growing. 
The sort of sensory, motor and ''mentaP tests that are used to 
measure physical and intellectual prowess usually give their 
best values in the neighbourhood of the age of twenty-five, or 
later. Usually, but not always: it is around the age often that 
hearing of very high-pitched sounds is most acute. ^ Information 
of this sort is intrinsically important, for it does something to 
confirm a theorem of wide significance which many clinicians 
have long taken for granted — that the time of onset and rate of 
ageing of the faculties and organs may vary independently 
within fairly wide limits. Other evidence tells against it. One of 
the most useful lessons to be learnt from the natural historian"'s 
studies of animal longevity^ is that the life span varies greatly 
in length between quite closely related types of organism. What 
can this mean, if not that the ageing process in the individual 
as a whole is geared by one or two limiting or '"master"' factors? 

Minot ""s special theory of the ageing process is just as un- 
usual as are his general laws, for he believed that cellular 

1 P. B. Medawar, Proceedings of the Royal Society. Series B, 132, p. 133, 

2 Y. Koga and G. M. Morant, Biometrika, 15, p. 348, 1924. Cf. the data 
summarized by V. Korenchevsky: Annals of Eugenics, 11, p. 314, 1942. 

^ The most important of these are by S. S. Flower. See note 2, p. 32. 



differentiation is the cause of the progressive fall away of 
growth potential. Cellular differentiation — the degree of muscli- 
ness of a muscle fibre, for example — has never been measured, 
but Minot guessed that if such a measurement were to be 
made, the curve of increasing differentiation would be found 
to be the exact complement of that which plots the declining 
energies of growth. To put it in another way: that which we 
call '"development'' when looked at from the birth end of life 
becomes senescence when looked at from its close. It is an 
attractive idea, but such little evidence as we have speaks 
against it. The tissue cultivator, who grows cells in blood and 
tissue media outside the body, finds that ""old"* cells have just as 
high a capacity for growth as young ones. They simply take a 
longer time to set about it.^ It is perfectly true that some very 
highly differentiated cells, like those of nerve and muscle, lose 
their power to multiply by fission. But that is more of a mech- 
anical accident than a slur upon their vitality; after all, a nerve 
cell may be some yards long. Neither adult nerve nor adult 
muscle has lost the power to grow^ and if a muscle or nerve 
fibre is cut into two, healing and replacement will start up from 
one end or the other. But whatever the rights and wrongs of 
Minofs special theory, he has left us with two ideas which 
any future theory of the ageing process must analyse and 
suitably explain: the first is the continuity of the ageing pro- 
cess, the second its great span in time. 

Some mention must now be made of the celebrated and 
widely misinterpreted views of Elie Metchnikoff on ageing.^ 
Metchnikoff believed that much of what in ageing seems to us 
to be very '"naturaP is in fact abnormal. How much of ageing 
he held to be so is far from clear, though he seemed to think, 
as Buffon did and later Flourens, that an animaPs total span 

^ Cf. the evidence summarized by P. B. Medawar, Proceedings of the 
Royal Society of Medicine^ 35, p. 690, 1942. 

2 The Prolongation of Life (trans. P. Chalmers Mitchell), London, 1910. 



of life should be between five and seven times the period that 
passes between birth and the onset of sexual maturity. Self- 
intoxication by the products of bacterial decomposition in the 
large intestine was chiefly to blame for the pathological changes 
of senescence. The theory has a homely origin. The mammals, 
Metchnikoff argued, do not void their faeces on the run, and 
yet are exposed to countless dangers by doing so when standing 
still. In order to choose the most appropriate time for defaeca- 
tion, mammals must therefore have large intestines in which 
to store their faeces.^ Bacteria flourish in the store-house so 
provided, and the absorption of their evil humours brings 
about a state that ranges from the malaise of constipation to the 
chronic and cumulative toxaemia of pathological senility. Cells 
intoxicated beyond redemption are attacked and eaten up by 
the phagocytic cells which, conveniently enough, Metchnikofl" 
himself had earlier discovered. 

Most laymen are convinced that there is something in this 
theory, and it has not lacked zoological champions of the 
greatest eminence. ^Certain it is,"* said AlacBride^ some twenty 
years later, in the course of a violent attack on mathematical 
biology, '"certain it is that in human beings, when the toxins 
produced by proteolytic enzymes are got rid of, many of the 
signs of old age may disappear."* But a biologist can pick holes 
in each single theorem. Some mammals do defaecate while 
running. The malaise of constipation is at once relieved by 
bowel movement, and fishermen who habitually defaecate at 
ten-day intervals are not the debile wrecks that Metchnikoff 's 
theory would have us think them. The large intestine, too, is 

^ It is a popular fallacy that faeces await evacuation in the rectum. This 
is so only in cases of chronic constipation. Cf. Sir A. Hurst, Proceedings of 
the Royal Society of Medicine, 36, p. 639, 1943. 

2 In the discussion of G. P. Bidder's Linnean Society lecture on ageing 
(note 2, p. 31). MacBride had been particularly upset by Karl Pearson's 
statement that mental deterioration in man began at the age of twenty- 



no mere dustbin. Herbivorous animals get some of their food 
from the action of cellulose-spHtting bacteria within it. The 
bacteria may, moreover, synthesize vitamins, which are ab- 
sorbed directly or may be recovered by eating the droppings 
themselves — a slap in the eye for Metchnikoffs theory. The 
theory is dead, and nothing is to be gained by propping it up 
into a sitting position.* 

In the first twenty years of this century, there began to 
accumulate new empirical evidence concerning the ''immor- 
tality'* of the ordinary non-reproductive cells of the body — 
more exactly, the immortality of the cell-lineages to which, by 
successive acts of fission, such cells may be ancestral. Leo Loeb 
and later, more clearly, Jensen showed that several tumours 
will grow indefinitely if handed on by grafting from one animal 
to another. 1 It used to be possible to buy from the laboratories 
of the Imperial Cancer Research Fund a rat bearing Jensen'*s 
rat sarcoma. Its cells are lineal descendants of those which 
Jensen first transplanted some forty years ago. The technique 
of growing cells outside the body proved as much for the cells 
of normal tissue. A strain of connective-tissue cells was started 
by Carrel and Burrows in 1912.2 The first year's growth was 
not enough to demonstrate the perpetuity of the cell lineage. 
We are 'not justified', said Ross Harrison in 1913, 'in referring 
to the cells as potentially immortal . . . until we are able to keep 
the cellular elements alive in cultures for a period exceeding 
the duration of life of the organism from which they are taken. 
There is at present no reason to suppose that this cannot be 

1 A clear elementary account of this early work is to be found in W. H. 
Woglom, Fifth Scientific Report of the Imperial Cancer Research Fund, 
p. 43, London, 1912. 

2 There are quite a number of popular accounts of this work, e.g. in 
A. Carrel, Man the Unknown, New York, 1935; L. du Nouy, Biological 
Time, London, 1936. 

* [A comparatively recent paper on Metchnikoff's theory is that by 
S. Orla-Jensen, E. Olsen and T. Geill, Journal of Gerontology, 4, p. 6, 1948.] 



done, but it simply has not been done as yet.' In due course it 
was done, and the strain was with us until 1939. Tissue-culture 
has other evidence to offer us of death. We are told that one 
of the last experiments of Thomas Strangeways was to cultivate 
the connective-tissue cells surviving in a sausage — as neat a 
demonstration as one could wish of the tenacity of the vita 
propria and the half-truth that is legal death. So let us submit 
yet another zoological simile of common speech to the censor- 
ship of our new wisdom. The earth stirs over MendePs grave 
when we say that two people are as like as two peas. Many fish, 
moreover, never drink. ''As dead as mutton"* is likewise super- 
annuated by the march of time; and those whose most pressing 
fear it is that they will be lowered living into their graves can 
have their doubts resolved: they will be. 

(The so-called '"immortahty" of the Protozoa is like that of 
the tissues: not an immortality of cells but an indeterminate- 
ness of cell lineages. Obviously the cell lineages of protozoa are 
in some cases immortal or indeterminate, for otherwise they 
could hardly be with us to-day. But does this immortality 
depend upon the performance of an occasional act of nuclear 
reconstitution, or can protozoa thrive for ever by the mere act 
of dividing asexually into two? The matter has long been 
controversial.^ Some of the early investigators believed that, in 
default of such ^rejuvenation"*, a protozoan lineage must under- 
go a microcosmic cycle of growth, maturity, decay and death, 
exactly like the cell population of higher organisms. Others 
believed that vegetative fission would suffice. When it came to 
be known that the former opinion was founded at least in part 
on the use of faulty techniques of cultivation, the latter dis- 
possessed it. But Jennings^ is now inclined to doubt whether 
asexual fission is in itself enough, and the more recent genetic 

^ Cf. H. S. Jennings, Problems of Ageing, ed. E. V. Cowdry, 2nd ed., 
pp. 24-46, Baltimore, 1942. 

2 Journal of Experimental Zoology, 99, p. 15, 1945. 



evidence suggests that some sort of nuclear rehabilitation is 
from time to time required. Ordinary asexual fission is, from 
the mechanic'o of the process, a very exact division of the parent 
organism into equal parts. The genetical sins of the parents — 
the lethal or unwholesome mutant genes — are thus allotted to 
their progeny with biblical justice and more than biblical pre- 
cision. The nuclear reconstitution spoken of above is, in effect, 
a device by which such genes may be eliminated from the 
stock. The organisms which inherit them die soon, or fail to 
reproduce; the others, often a minority, carry on.)^ 

With such new facts as these at his disposal, and others of 
great value added by himself, Raymond PearP made the next 
important attack on death in 1922. Pearl himself showed that 
an animaPs span of life was governed by inherited factors and 
was within certain limits subject to experimental modification. 
The total span of life may be increased not by adding a few 
extra years to its latter end nor, if it comes to that, by inter- 
calating new life at any intermediate period, but rather by 
stretching out the whole life span symmetrically, as if the seven 
ages of man were marked out on a piece of rubber and then 
stretched. The length of life may thus be treated as a function 
of the rate of living. One simple way of lowering the rate of 
living — an ingredient of many a centenarian''s recipe for long 
life — is to withhold with known precision the sort of food that 
is used for the supply of energy: a restriction of calories, as we 
say, rather than a systematic maZnutrition. McCay and his 
colleagues^ have shown that by such means the life span of rats 
may be greatly lengthened. The same is true of flatworms, as 

1 Cf. B. F. Pierson, Biological Bulletin, 74, p. 235, 1938; T. M. Sonneborn, 
ihid., p. 76. 1 am obliged to Professor J. B. S. Haldane for pointing out the 
significance of their evidence. [For more recent evidence on senescence in 
protozoa, see Alex Comfort, The Biology of Senescence, London, 1956.] 

^ The Biology of Death (Lowell Lectures), Philadelphia, 1922; The Rate 
of Living, London, 1928. 

3 Cf. C. M. McCay, pp. 680-720, in Problems of Ageing (note 1, p. 27). 



Child told us;i of certain sea-squirts, and of the aberrant, worm- 
like creatures known as Nemertines.^ These latter have the 
advantage of the rat, for if deprived of food they react by- 
growing smaller, thus literally retreating into second childhood. 
They do not quite exactly retrace their steps, ''advancing back- 
wards'* (as was said of a recent famous military campaign) along 
the path they followed in development; but in a sense they 
cheat Time. The fact that starved rats outlive those which 
habitually eat sufficient is often used as evidence of the rela- 
tivity of biological time; but in reality, it is evidence less of the 
tortuous mysteries of time and space than of the virtues of 
sobriety and moderation. 

In the extreme case, when life is held altogether in abeyance, 
we may properly speak of immortality. Freeze a tissue such as 
mammalian skin to the temperature of liquid air (something 
less cold will do) and the resumption of life will then await the 
convenience of the experimenter.^ The idea is an old one. 
Until he tried to freeze two carp, John Hunter — * 

'imagined that it might be possible to prolong life to any period 
by freezing a person. ... I thought that if a man would give up 
the last ten years of his life to this kind of alternate oblivion and 
action, it might be prolonged to a thousand years; and by getting 
himself thawed every hundred years, he might learn what had 
happened during his frozen condition. Like other schemers, I 
thought I should make my fortune by it; but this experiment 
undeceived me.' 

1 C. M. Child, Senescence and Rejuvenescence, Chicago, 1915. 

2 See J. Needham, Biochemistry and Morphogenesis, pp. 524-9, Cam- 
bridge, 1942. 

3 Cf. R. Briggs and L. Jund, Anatomical Record, 89, p. 75, 1944; J, P. 
Webster, Annals of Surgery, 120, p. 431, 1944. The author has often 
confirmed their observations. [See R. E. Billingham and P. B. Medawar, 
Journal of Experimental Biology, 29, p. 454, 1952.] 

* J. Hunter, Of the Heat of Animals, in The Works of John Hunter, 
F.R.S., ed. J. F. Palmer, Vol. 1, p. 284. The phenomenon which Hunter 
unluckily failed to demonstrate has been called 'anabiosis'. 



These particular carp died, though latter-day experimenters 
have been more lucky.^ 

Raymond Pearl agreed with Weismann that in some manner 
or other natural death had evolved, but that it evolved under 
the auspices of natural selection he irritably denied. (''Probably 
no more perverse extension of the theory than this was ever 
made."*) Yet for so brilliant a man, PearPs own theory of the 
mechanism of ageing in the individual is curiously inadequate. 
""Specialization of structure and function necessarily makes the 
several parts of the body mutually dependent for their life upon 
each other. If one organ or group, for any accidental reason, 
begins to function abnormally and finally breaks down, the 
balance of the whole is upset and death eventually follows. '' 
But is not this a description of the ""proximate cause** of almost 
any form of death? Something gives way, no doubt: one man 
will be as old as his arteries, another as his liver. But gross 
abnormality apart, why should any organ break down? Appar- 
ently because of the wear and tear of merely working, and 
Pearl tells us that ""those organ systems that have evolved 
farthest away from the original primitive conditions . . . wear 
longest under the strain of functioning\ It is only towards the 
end of his book that Pearl puts forward his theory in this 
relatively specific form. Earlier — and see how much more easily 
he breathes the air of amorphous generalization — he tells us 
that the somatic death of higher organisms ""is simply the price 
they pay for the privilege of enjoying those higher specializa- 
tions of structure and function which have been added on as a 
sideline to the main business of living things, which is to pass 
on in unbroken continuity the never-dimmed fire of life itself \ 

^ E.g. N. A. Borodin, Zoologische Jahrbuch, 53, p. 318, 1934. [To-day, 
thanks to the work of R. Andjus of the University of Belgrade, even 
mammals can survive being frozen: the subject of 'hypothermia' and of 
tissue storage by freezing has been admirably reviewed by R. E. Billing- 
ham in New Biology, 18, p. 72 (London: Penguin Books, 1955). See also 
A. U. Smith, J. E, Lovelock and A. S. Parkes, Nature, 173, p. 1136, 1954.] 



A stirring thought; but Johannes Muller had said as much 
some eighty years beforehand^ and with proper scientific 
caution had remarked: ''This has the appearance of explaining 
the phenomena, but is in reahty a mere statement of their 
connection, and it is not even certain that as such it is 

Let us now turn to one last famous speculation on the pro- 
blem of natural death. Minot, we saw, left us with the capacity 
for growth as an upside-down measure of the rate of ageing. 
Suppose an animal increased in size indefinitely: would it die a 
natural death? Hardly, if so important a function as growth 
were left undimmed by age. But before hearing Bidder''s 
answer,^ the question can be put a little more exactly. The 
distinction is not between animals which continue to grow and 
animals which stop growing but between animals without and 
with a limit to their size. How the limit is approached is neither 
here nor there. It may be approached asymptotically, as in 
mathematical theory, or finally — to a maximum — as for all 
intents and purposes it is in fact. According to Bidder, fish 
grow without limit and never undergo senescence nor suffer 
natural death. Indeed, he does not 'remember any evidence of 
a marine animal dying a natural death'. Now a mechanical 
limit is set to the size of animals on land, as Galileo and many 
others since have taught us; and according to Bidder this limit 
is set, or has come to be set, by an intrinsic limitation of the 
power of growth, with senescence as its outcome. ''Did old age 
and death only become the necessary fate for plants and 
animals when they left the swamps, claimed the land, and 
attempted swiftness and tallness in a medium -^^ of their 
specific gravity?"* Bidder believes that this is so, if the quite 

J. Muller, Elements of Physiology (trans. W. Baly), Vol. 1, pp. 36-6 
(and cf. Vol. 2, p. 1660), London, 1840-2. 

2 G. P. Bidder, Proceedings of the Linnean Society ^ p. 17, 1932, British 
Medical Journal, 2, p. 583, 1932. 



special category of '"parentaP death, like that suffered by the 
male salmon, is left out of count. 

We will skip blindfold over the causal nexus that relates the 
limitation of growth to the degenerative changes of old age, 
and ask ourselves if Bidder"'s main thesis, that marine animals 
do not die natural deaths, is in fact true. It is a 'highly debat- 
able problem** — that is to say, one with so little evidence to its 
credit that no debate is in reality worth while. We have, it 
appears, little to say about the death offish that Ray Lankester 
did not say in his Prize Essay on longevity some eighty years 
ago:^ ''they are not known to get feeble as they grow old, and 
many are known not to get feebler\ ""Real evidence is practically 
non-existent,"* said Major Flower,^ though he could tell us that 
'under favourable circumstances some fresh-water fishes may 
live for half a century\ The fact of the matter is that the energy 
that might have been devoted to a theoretically straightfor- 
ward solution of the problem has very often been dissipated in 
digging up anecdotes about longevity from obsolete works of 
natural history. Nor has the research been theoretically 
prudent, for often no distinction has been made (though 
Lankester insisted on it) between the mean expectation of life 
and the total life span. It proves that we cannot accept the 
claims of most of the famous human more-than-centenarians, 
so what faith are we to have in the pedigrees of tortoises and 
carp? No one has yet made a systematic study of whether even 
mammals in their natural habitat do indeed live long enough to 

1 E. Ray Lankester, On Comparative Longevity on Man and the Lower 
Animals, London, 1870. 

2 See the series of articles in the Proceedings of the Zoological Society 
(latterly Series A), 1925, 1981, 1935-8. [Flower's last paper, published post- 
humously, on the alleged longevity of elephants, should on no account be 
missed: see the Proceedings of the Zoological Society, 117, p. 680, 1947. 
The old original Jumbo ('Old Jumbo carried generations of London children 
round the zoo in Regent's Park') died at 24, Alice at 50, and Napoleon's 
Elephant at about 63. Flower dissects the legends of their longevity with 
admirable skill.] 



reach a moderate though certifiable degree of seniHty. ^As a 
matter of fact, the contribution that senescence makes to 
accidental death can be deduced with reasonable accuracy 
from the mathematical character of the actuary's life table. 
For if the '"force of mortality** were constant and independent 
of age; if, that is to say, the chances of dying were the same in 
the age interval 100-101 years as in the interval 10-11 years; 
then the curve defined by the life table would be of the familiar 
die-away type that describes, for example, the loss of heat from 
a cooling body. But no life table has yet been made for a 
mammalian species in the wild. All that can be said so far, in 
the spirit of Lankester'*s generalization, is that some mammals 
do not appear to live that long. Hinton''s studies^ on fossil and 
recent voles of the genus Arvicola showed that ""not only are the 
molars still in vigorous growth, but the epiphyses of the limb 
bones are still unfused with their shafts. Apparently, that is as 
far as actual observation goes, voles of this genus are animals 
that never stop growing and never grow old. But no doubt, if 
one could keep the vole alive in natural conditions, but secure 
from the fatal stroke of accident, a time would come when . . . 
the animal would become senile and die in the normal manner.' 
Burt's study^ of mice of the genus Peromyscus led to a similar 
conclusion; but there, so far as I know, the matter stands. The 
difficulties of constructing life tables for animals in the wild are 
technically formidable, but they must be solved.* 

From the standpoint of evolutionary biology an animal's 
expectation of life in its natural surroundings is much more 
significant than the degree of decrepitude to which it may be 
nursed in laboratory or zoo. It is a fair guess that much of what 

1 M. A. C. Hinton, Monograph of the Voles and Lemmings, Vol. 1, p. 48, 
British Museum, London, 1926. 

2 W. H. Burt, No. 48 in Miscellaneous Publications of the Michigan 
University Museum of Zoology, May, 1940. I must thank Mr D. Chitty 
for this reference. 

* [As they are beginning to be: see the literature cited in note 1, p. 22.] 

c 33 



we call the senile state is in the ecologist''s sense merely patho- i 
logical. Senility is an artifact of domestication, something dis- \ 
covered and revealed only by the experiment of shielding an 
animal from its natural predators and the everyday hazards of ' 
its existence. In this sense, no form of death is less ""naturar '' 
than that which is commonly so called. ] 

Some interesting conclusions may be drawn from the fact I 
that the latter end of life is ecologically atrophic or vestigial. i 
It has several times been pointed out^ that the changes which j 
an animal may undergo after it has ceased to reproduce are 
never directly relevant, and are in most cases quite irrelevant, 
to the course of its evolution. A genetic catastrophe that befell 
a mouse on the day it weaned its last litter would from the 
evolutionary point of view be null and void. This state of 
affairs is tacitly acknowledged in the celebrated half-truth that 
'parasites live only to reproduce"*: the whole truth is that what 
parasites do after they reproduce is not on the agenda of 
evolution. The same applies to what may befall a mouse when it 
reaches the age of three, though in fact it never (or hardly ever) 
lives that long. We shall return to this point later. For the 
present it may be said that the existence of a post-reproductive 
phase of life is not causally relevant to the problem of ageing, 
for it is just that very ingredient of the ageing process — the 
decline and eventual loss of fertility — which it is our chief 
business to explain. 

What is the upshot of all this speculation? I think many 
biologists would agree that Weismann was in principle correct, 
and that the process of senescence in the individual and the 
form of the age-frequency distribution of death that mirrors 
it statistically have been shaped by the forces of natural 

^ Cf. G. G. Simpson, Tempo and Mode in Evolution^ p. 183, Columbia 
University Press, 1944. 



selection.* But before looking into this belief more closely, 
it will be as well to start this section, like its less technical 
predecessor, with a few definitions. 

First, ''evolution\ Biologists often speak of organs, tissues 
and even cells *"evolving% but it must be recognized that this 
manner of formulation is by modern lights imprecise, or, what 
is not quite the same thing, inexpedient. These various things 
do indeed participate in evolution, just as our noses participate 
in our motion without themselves being mobile. What moves in 
evolution, what evolves, is an animal population, not an indi- 
vidual animal; and the changes that occur in the course of 
evolution, if we put a magnifying glass to them instead of 
' feeling obliged to peer dimly down the ages of geological time, 
are changes in the composition of a population and not, prim- 
arily, in the properties of an individual. In visual analogy they 
are to be likened, not to a transformation scene at the panto- 
mime, but to the sort of overlapping transformation we watch 
at the cinema when one ""sef slowly evaporates and is dis- 
possessed of the screen by another. 

Further, whatever form evolution may take, or whatever 
may bring it about, contributions to evolutionary change are 
paid, if they are paid at all, in one currency alone: offspring. 
Animals favoured by the process which, wise after the event, 
we call ^'natural selection"*, pay an extra contribution, however 
small, to the ancestry of future generations; and this brings 
about just that shift in the genetical composition of a popula- 
tion which we call an '"evolutionary change\ The problem of 
measuring natural selection, which so worried Karl Pearson,^ 
is thus solved: the magnitude of natural selection is measured 
by the relative increase or decrease in the frequency with 

1 Cf. K. Pearson, The Chances of Death and Other Studies in Evolution, 
2 vols., London, 1897. 

* [The argument sketched in this section is developed more fully in the 
essay which follows, An Unsolved Problem of Biology.^ 



which the factor which governs some heritable endowment 
appears in the population. 

I said, earlier on, that any theory of the origin of the ageing 
process must take two things into account: the early onset of 
what is in the technical sense senescence, and the continuity 
of its expression through life. I would like now to suggest that 
the 'force of mortality"* has been moulded by a physical oper- 
ator that has the dimensions of time x luck. Let us examine 
how natural selection will work upon a population that is 
potentially immortal; of which the individuals remain, for all 
the time that they are alive, in the fullness of physical maturity. 
Such a population will contain old animals and young. The old 
are old in years alone: we are so used to hearing the overtones 
of senility in the word 'old"* that we must forcibly adjust our- 
selves to accept this important qualification. The old animals 
I shall speak about are '"in themselves"* (to use a category of lay 
diagnosis) '■young\ They will no doubt have the advantage of 
their juniors in reflex and immunological wisdom, but these 
advantages will in the first approximation be disregarded. 

Upon this population exempt from age decline we shall now 
superimpose a variety of causes of death that are wholly 
random or haphazard in their manner of incidence. The causes 
of death being random in nature, and susceptibility to it inde- 
pendent of age, it follows that the probability that an animal 
alive at the beginning of any span of time will die within its 
compass is likewise constant. The one-year-old is just as likely 
to see his second birthday as is the fifty-year-old to see his 
fifty-first. But the chances at birth of living to age 1 and age 50 
are very diff'erent indeed; for as Weismann pointed out, though 
the significance of it escaped him, the older an animal becomes 
the more frequently is it exposed to the hazard of random 
extinction. Likewise a coin that has turned up heads ten times 
running will turn up heads on the eleventh spin in just 50 per 
cent, of trials; but the chances of turning up heads eleven times 



running are very small indeed. The upshot of this is that young 
animals will always outnumber old. 

Let us in imagination mark a group of 100,000 animals at 
birth and follow it through life, supposing that the chance of 
dying within any small interval of time is constant, and equal 
to one-tenth per annum of those that remain alive to submit 
to the hazard. The survivors at the end of the first year will be 
90,000; at the end of the second year, nine-tenths of those alive 
at its beginning, namely 81,000; and so on, through 72,900, to 
numbers which obviously get very small. In a population with a 
4ife table"* such as this, supposing that it is not decreasing in 
numbers, a certain steady state of ages will be reached, a 
certain definite age-spectrum or composition with regard to 
age. At this steady stage, youngsters are being fed into the 
lower reaches of each age-group at the same rate as death and 
the passage of time remove them from it. The shape of this 
""stable age distribution"* (which is moulded, odd though it may 
seem, by the birth-rate per head alone) is that of a die-away 
exponential curve, such as one so often meets in the numerical 
treatment of natural data. The number of animals in each 
age-group bears a constant ratio, greater than unity, to the 
number of animals in the age-group following next. 

What is important from our point of view is that the con- 
tribution which each age-class makes to the ancestry of future 
generations decreases with age. Not because its members 
become progressively less fertile; on the contrary, it is one of 
our axioms that fertility remains unchanged, so that the repro- 
ductive value per head is constant;^ but simply because, as age 
increases, so the number of heads to be counted in each age- 
group progressively falls. It is at least as good a guess as 
Weismann made, that the process of senescence has been 
genetically moulded to a pattern set by the properties of this 

^ The term is teclmically defined in R. A. Fisher, The Genetical Theory 
of Natural Selection, Chap. 2, Oxford, 1930. 



'immortar age distribution. It is by no means difficult to 
imagine a genetic endowment which can favour young animals 
only at the expense of their elders; or rather, at their own ex- 
pense when they themselves grow old. A gene or combination 
of genes that promotes this state of affairs will under certain 
numerically definable conditions spread through a population 
simply because the younger animals it favours have, as a 
group, a relatively large contribution to make to the ancestry 
of the future population. It is far otherwise with a genetic 
endowment which favours older animals at the expense of 
young. Reflection will show that the gene or genes concerned 
cannot plead for a retrospective judgement in their favour; for 
before the animals which bear these genes give outward *'pheno- 
typic"* evidence of the fact, they are on equal terms with those 
that do not. The greater part of the ancestry of the future 
population will thus have been credited indifferently to both 
types, because a gene qualifies for the preferential action of 
natural selection only when, to put it crudely, it manifestly 
works. This does not imply that a late-acting gene which 
confers selective advantage cannot spread through the 
population. It can indeed do so; but very much more slowly 
than a gene which gives evidence of itself earlier on. The 
later the time in life at which it appears, the slower will be 
its rate of spread; and the rate in the end becomes vanishingly 

The consequence of any decline in the fertility of older 
animals is cumulative. Once it has happened, a new set of 
events may be put in train. Consider the fate of genetic factors 
that make themselves manifest in animals that bear them, not 
at birth nor in the first few days of life but at some time later 
on. Quite a number of such genes are known, and what is said 
of them applies equally to genes which have an expression, 
but a variable form of expression, throughout the whole span 
of life. It may be shown that if the time of action or rate of 



expression of such genes is itself genetically modifiable, then, 
if the gene confers selective advantage, its time of action or of 
optimal expression will be brought forward towards youth, as 
it spreads through the population. If, by contrast, the gene is 
'disadvantageous**, then its time of action or threshold of un- 
favourable expression will be pushed onwards in life while it is 
being eliminated from the population. The former process may 
be called a precession of favourable gene effects; the latter, a 
recession of unfavourable effects. Neither process can come into 
operation unless the fertility of the population declines with 
age, so that the reproductive value of its members falls; and 
the latter process, the recession of unfavourable gene effects, 
will be modified by the fact that the later an ""unfavourable"* 
gene comes into operation, the slower will be the process of its 
removal from the population. (At some critical late age, per- 
haps, an unfavourable gene is eliminated so slowly that natural 
selection cannot challenge its reintroduction into the popula- 
tion in the process of gene mutation.) The precession of 
'favourable'' gene effects will in its turn be modified by the fact 
that reproduction cannot start at birth, and nature has found 
in higher animals only the most indirect substitutes (maternal 
care, and the blunderbuss of huge fecunditv) for the theoretic- 
ally desirable state of affairs in which an animal is born 
mature. Because of the hazards to which baby animals are 
exposed (and this is just as true of human beings) the repro- 
ductive value of the individuals always rises to a maximum 
before eventually it falls; and it is at the epoch of this maxi- 
mum, therefore, that the 'precession"* of favourable gene effects 
will automatically come to halt. It is not surprising, then, to 
find that in human beings the 'force of mortality"* is lowest just 
when the reproductive value would in the members of a prim- 
itive society be highest — in the neighbourhood of the four- 
teenth or fifteenth years of life. ^ Nor is it surprising to find that 
^ A correlation pointed out by R. A. Fisher (see note 1, p. 37). 



'senescence** begins then, rather than at the conventionally 
accepted age of physical maturity somewhat later on. 

The foregoing paragraphs represent no more than a few 
extra guesses woven in among Weismann'*s original hypothesis 
of ageing. If what Weismann believed is true, then nothing very 
radical can be done by way of modifying the course of growing 
old. Scientific eugenics could in the long run give us a more 
generous span of life; but only, it seems, by engaging life in 
lower gear, by piecing out the burden of the years into a larger 
number of smaller parcels, so prolonging youth symmetrically 
with old age. But the inevitability of old age does not carry 
with it the implication that old age must be a period of feeble- 
ness and physical decay. If specific secretions of the ductless 
glands fail; if assimilation becomes less efficient, so that essential 
food factors fail to penetrate the gut wall; if chronic low-grade 
infections persist because the defences of the body lack power 
to overcome them; in all such cases it should be possible to 
remove, at least for a while, any ingredients of the senile state 
for which they may be specifically responsible. The solution of 
these problems is a matter of systematic empirical research. 

Side by side with research of this type there should be under- 
taken a thoroughgoing physiological analysis of the mechanism 
of ageing. I shall sketch one possible line of analysis here, 
because although the layman often understands the nature of 
scientific problems and can usually grasp the principles of their 
solution, he has, as a rule, very little idea of how scientific 
work is actually done. 

If a physiologist were to study the problem of ageing from 
scratch, he would not even begin to try to modify the time- 
course of senescence by the administration of vitamins or 
elixirs compounded of the juices of the glands. He would first 
of all try to piece together a full empirical description of the 



phenomenon of ageing, as it is reflected in structural changes 
of tissues and cells and, more particularly, in the type and in- 
tensity of tissue and cellular metabolism. Only scraps of such 
information are now available: he would have to collect more. 
(The physiologist might in any case become more fully aware 
of the dimension of time in his experimental work. Nearly all 
his work is done with mature animals; studies on youngsters 
and animals past the reproductive period are far too few.) 

With an adequate background of purely descriptive evid- 
ence, the physiologist could then bring the experimental 
method to bear. The first problem he would seek to solve is 
this: is the phenomenon of ageing something ''systemic*' in 
nature — something manifested only by systems of the degree of 
organization of whole animals — or is it intrinsically cellular? 
Studies on tissue cultivation have given a partial answer to this 
question, but there are grounds for supposing that in certain 
critical respects it is misleading. One promising alternative that 
has become available to him is the technique of tissue and 
oi^an transplantation between animals of different ages. The 
majority at least of the members of very highly inbred strains 
of mice are from the standpoint of tissue-interchange genetic- 
ally identical, for after many generations of repeated brother- 
to-sister mating they come to resemble each other (sexual 
differentiation apart) almost as closely as identical twins. One 
may therefore interchange parts of their bodies on a scale 
limited only by the exigencies of technique; one may make 
time-chimeras of youth and old age.* How, then, does tissue 
transplanted from a baby animal to a dotard develop in its 
''old'' environment? Does it rapidly mature and age, or does it 
remain like a new patch on an old pair of socks? Conversely, 

* [The grafting of tissues between animals of different ages might be 
described as 'heterochronic', and it does for age or time what 'heterotopic' 
grafting does for place or space; see The Uniqueness of the Individual. 
Professor P. L. Krohn is making a particular study of these problems.] 



what is the fate of tissue grafted from old animals into young- 
sters? If ordinary laboratory mice are used for such experiments, 
as very likely they have been, or even what are sometimes with 
undue optimism called ''pure strains', then the evidence is 
falsified at the outset; for the transplantation of tissues be- 
tween animals very little dissimilar genetically simply provokes 
an immunity reaction, not different in principle from that which 
governs the outcome of certain blood transfusions, as a con- 
sequence of which the grafted tissue is destroyed."'^ But if suit- 
able genetic precautions are taken, these problems and others 
of equally wide compass are capable of solution. Only when 
they are solved can the physiologist begin to ask more specific 
questions, such as whether the determinative factors of ageing 
are humoral in nature or of some other more complex type. 

It is rather urgent that research of this type should be under- 
taken. Man''s mean expectation of life at birth has increased 
very dramatically over the last 100 years, but chiefly as a 
consequence of reduced mortality in infancy and childhood. 
The mean expectation of life at the age of forty has increased 
hardly at all. But because of this reservation for life of many 
who would otherwise have died, the age-spectrum of the popu- 
lation, i.e. the proportion of its members within each age-group 
of life, is in many civilized countries shifting slowly towards old 
age.f In forty years'* time we are to be the victims of at least a 

* [See The Uniqueness of the Individual herein.] 

t For the population of the U.S., see W. S. Thompson and P. K. Whelp- 
ton, Estimates of Future Population of the United States, 1940-2000, 
National Resources Planning Board, Washington, 1943. F. W. Notestein 
et al. The Future Population of Europe and the Soviet Union, Population 
Projections 1940-1970, Office of Population Research, Princeton University, 
1944. [For the population of the U.K., see Reports and Selected Papers of 
the Statistics Committee (\^ol. 2 of Papers of the Royal Commission on 
Population), London, H.M.S.O., 1950. Secular changes in the age-structure 
of populations since demographic data first began to be compiled are 
summarized by L. I. Dublin, A. J. Lotka and M. Spiegebnan, Length of 
Life: a Study of the Life Table, New York, 1949.] 



numerical tyranny of greybeards — a matter which does not 
worry me personally, since I rather hope to be among their 
numbers. The moral is that the problem of doing something 
about old age becomes slowly but progressively more urgent. 
Something must be done, if it is not to be said that killing people 
painlessly at the age of seventy is, after all, a real kindness. 
Those who argue that our concern is with the preservation of 
life in infancy and youth, so that pediatrics must forever take 
precedence of what people are beginning to call "'gerontology'', 
fail to realize that the outcome of pediatrics is to preserve the 
young for an old age that is grudged them. There is no sense 
in that sort of discrimination. 



An Unsolved Problem 
of Biology 

The problem I propose to discuss is that of the origin and 
evolution of what is commonly spoken of as *'ageing\^ It is a 
problem of conspicuous sociological importance. Everyone now 
knows that the proportion of older people in our population is 
progressively increasing, that the centre of gravity of the 
population is shifting steadily towards old age. Using a plaus- 
ible combination of hypotheses, one among several, the Stat- 
istics Committee of the Royal Commission on Population 
predicts that in half-a-century''s time one-quarter of our popu- 
lation will be not less than sixty years of age. The economic 
consequences of such an age-structure are all too obvious. Now 
biological research is by no means uninfluenced by the econ- 
omic importunities of the times, and there can be little doubt 
that the newly awakened interest of biologists in ageing — or 
the hard cash that makes it possible for them to gratify it — is a 
direct reaction to this economic goad. Unfortunately, scientists 

^ [The preamble appropriate to an Inaugural Lecture has been left out.] 
I have kept closely to my lecture as it was actually delivered, except that 
(a) I have left out an argument which, on further reflection, seems much 
less relevant and convincing than I formerly believed it, and {b) I have 
tried to answer in footnotes some particularly cogent criticisms by my 
colleagues. I have had the good fortune to consult with Professor L. S. 
Penrose on certain problems relating to the action of natural selection on 
human beings, and have had the most valuable advice from Professor 
J. B. S. Haldane, some of whose ideas are presented here as if they were 
my own. 



have been slow to realize that the biologically important con- 
sequences of this secular increase in average longevity began 
to be apparent three-quarters of a century ago and are now on 
the threshold of completion. About seventy-five years ago, the 
mean expectation of life at birth in England and Wales began 
to exceed, as it now greatly exceeds, the age beyond which 
child-bearing virtually ceases. Women have had nearly all their 
children by the time they are forty-five, but may novv expect, 
on the average, to live some quarter of a century longer. The 
fertility of men lasts beyond that of women and ends less 
sharply, but, roughly speaking, three-quarters of the male 
population is still alive at an age at which it can be credited 
with 99 per cent, of its children. The principal causes of death 
have changed accordingly. Fifty years ago the major killing 
diseases were pneumonia and tuberculosis, both of infective 
origin; to-day they are cancer and what is compendiously called 
cardiovascular disease. Susceptibility to both cancer and the 
cardiovascular diseases is in some degree influenced by heredity, 
and should therefore be subject to those forces, of ''natural 
selection**, that discriminate between the better and the genet- 
ically less well endowed. (To speak of '"discrimination"* is, of 
course, to put the matter in too literary a way; let us say that 
people with different hereditary endowments do not have 
children in strict proportion to their numbers; some of them 
take more than their numerically fair share of the ancestry of 
future generations.) But cancer and the cardiovascular diseases 
are affections of middle and later life. Most people will already 
have had their children before the onset of these diseases can 
influence their candidature for selection. In the post-repro- 
ductive period of life, the direct influence of natural selection 
has been reduced to zero^ and the principal causes of death 
to-day lie just beyond its grasp. 

^ The word 'direct' is important. Grandparents, though no longer fertile, 
may yet promote (or impede) the welfare of their grandchildren, and so 



How it is that the force of natural selection becomes attenu- 
ated with increasing age I hope to explain very fully later. 
What is important in the meantime is that one should realize 
how, in the last seventy-five years, the whole pattern of the 
incidence of selective forces on civilized human beings has 
altered. We are not now waiting for our ageing population to 
produce biological changes of first-class importance, as some 
demographers seem to suggest. The changes have already 
happened. We have already entered a new era in the biological 
history of the human race. 


It is a curious thing that there is no word in the English 
language that stands for the mere increase of years; that is, for 
ageing silenced of its overtones of increasing deterioration and 
decay. At present we are obliged to say that Dorian Gray did 
not exactly 'age% though to admit that he certainly grew older. 
We obviously need a word for mere ageing, and I propose to use 
''ageing'' itself for just that purpose. 'Ageing' hereafter stands 
for mere ageing, and has no other innuendo. I shall use the 
word ''senescence'' to mean ageing accompanied by that decline 
of bodily faculties and sensibilities and energies which ageing 
colloquially entails. Dorian Gray aged, but only his portrait 

influence the mode of propagation of their genes. A gene for grandmotherly 
indulgence should therefore prevail over one for callous indifference, in 
spite of the fact that the gene is propagated per procurationem and not by 
the organism in which its developmental effect appears. Selection for 
grandmotherly indulgence I should describe as 'indirect', and the indirect 
action of selection becomes important whenever there is any high degree of 
social organization. The genes that make for efficient and industrious 
worker bees, for example, are of vital importance to the bee community, 
though not propagated by the worker bees themselves. Dr Kermack points 
out that the distinction between 'direct' and 'indirect' selection can easily 
be misleading, because in the outcome their effects are both the same. Let 
us admit, however, that there is a distinction of genetical procedure, though 
it might well have been embodied in better-chosen terms. 




disclosed the changes of senescence. I hope that makes it 

Senescence means a decline of vitality. How is this to be 
more precisely defined and measured? One may set about trying 
to measure senescence in two entirely different sorts of ways. 

The first sort of measure is personal, in the sense that it is 
carried out on individual animals. Quite a number of schemes j 

of measurement are at our disposal. For example, the rate at 
which wounds heal provides some sort of measure of what we i 

vaguely mean by vitality, since it depends on the multiplication I 


or migratory activity of cells. What sort of answer does it give? 
So far as we know, the answer is that the rate of wound healing 
is highest at birth and steadily declines thereafter. In terms of 
this measurement, therefore, senescence begins at birth and 
the *"prime of life"* is something of a fiction. Or we might reason- 
ably choose a measure founded on the acuity of the senses. 
The acoustical prime of life, for example, appears to be in the 
neighbourhood of the age often, for we are said to hear sounds 
of higher pitch at ten than earlier or thereafter. On the other 
hand physical strength, endurance and the niceties of muscular 
co-ordination reach their peak at about twenty- five. 

All these are very piecemeal measures. The best, perhaps, is 
that originally devised by Minot — the multiplicative power of 
the tissues of the body, that is, their capacity to increase by 
further growth in the manner in which they themselves were 
formed. Organisms tend to grow by compound interest, for that 
which is formed by growth is itself usually capable of further 
growing. But the rate of interest falls; the organism grows like 
a sum of money which, invested at birth at (say) 10 per cent, 
compound interest, gathers in a lower rate of interest year by 
year. The rate of interest does indeed fall from birth, and it is 
at birth, if Minot is to be believed, that senescence must be 
said to begin. And so, in some perfectly respectable sense, it 
does; but if we pursue this train of thought by asking in what 



manner the rate of interest falls, we shall be led by Minot into 
an attractive paradox. The answer is that from birth onwards 
the rate of interest falls steadily at a rate which itself steadily 
falls. Not only does senescence begin at birth, but it is going 
on much faster in the early years of life than latterly. The child 
is hurrying precipitately towards his grave; his elders, appro- 
priately enough, proceed there in a more decent and orderly 

None of these personal measures is of more than limited 
value. They are together incomplete, and severally give differ- 
ent answers; nor can they be made to add up to give a single 
figure that represents a measure of senescence in the round. 
Let us therefore turn to a scheme of measurement founded on 
wholly different principles. 


The second sort of measure is not personal, but statistical. 
We have agreed that senescence is a decline of what may be 
vaguely called vitality, and must now ask what property it is 
that changes as a direct outcome of that decline. The property 
is, in a word, vuhierability to all the mortal hazards of life; and 
it is measured by the likelihood of dying within any chosen 
interval of age. 

The measurement of vulnerability is in principle very easy. 
Imagine 100,000 animals, each of which is labelled or otherwise 
identified at birth and followed throughout its life; and suppose 
one keeps a record of the age at which each dies, keeping the 
record open until the death of the most long-lived. Such a 
record might well be called a Death Table, but, by an agreeable 
euphemism, it is in fact called a Life Table. If we plot the 
number of survivors against age, the curve so defined starts 
with age at 100,000 and falls to zero at the age of about 100 
years. Fig. 1 illustrates the shape of the life-table curve for 
human beings. 



From such a curve one may compute the death-rate at any 
age of Kfe, for that is simply its slope, the rate of decline of the 
number of survivors; the mean expectation of further life at birth 
or at any other age; and the likelihood at any one age of living 






75.000 - 



age in years 
Fig. 1 




to any other. The property that concerns us, however, is that 
which is called the specific death rate or, less aridly, the 'force 
of mortality\ the likelihood of dying within each interval of 
age. In a first approximation, which is all that is necessary for 
our purpose, the force of mortality is the quotient of this 

Number of organisms that die within any chosen interval of age 
Number of organisms alive at the beginning of the interval 

If, for example, 100 men reach age eighty-nine, but only 80 
of them reach age ninety, then the force of mortality in the 
ninetieth year of life is simply 0.2 (20 per cent., or 200 in every 
1000). If there is no senescence in the population — if vitality 
does not decline, so that there is no greater likelihood of dj'ing 
at any one age than at any other — then the force of mortality 
must necessarily be constant. Its members die, to be sure; but 

D 49 


a man who has just celebrated his eightieth birthday anni- 
versary is no more or less likely to celebrate his eighty-first than 
is a seven-year-old to celebrate his eighth. In my diagram, the 
force of mortality, being constant, would appear as a straight 
line parallel to the axis defining age. 

In real life it is far otherwise. As Fig. 2 shows, the vulner- 
ability of newborns is, not unexpectedly, very high; not until 
nearly the seventieth year of life does it become so high again. 
The curve of the force of mortality falls precipitously to a 
minimum around age twelve and then climbs upwards, slowly 
at first and latterly much faster. Age twelve (or thereabouts) 
is therefore the actuarial prime of life; at twelve one is more 
likely than at any other age to survive one further year, or 
month, or minute. But notice the smoothness of the curve that 
defines the force of mortality in later life. There is no break or 
singularity to give evidence that at any later age development 
and maturation are at least completed and that deterioration 
then sets in. Any complete theory of the origin and evolution of 
senescence must explain the smoothness and coherence of the 
curve of increasing vulnerability. It is not quite good enough 
merely to think up reasons why very elderly animals should 

Because there are clearly special reasons why baby animals 
should be more vulnerable, though no less charged with vitality 
than their elders, I am proposing to neglect the arc of the curve 
of the force of mortality that lies before its minimum, but to 
use its later stretch as a measure of the degree of senility. This 
is a decision that cries aloud for qualifications and reservations, 
and it is part of my purpose to reveal what some of these 
may be. 

You will notice first that although the force of mortality may 
purport to measure a process that happens in the life of an 
individual animal (decline of "vitality", or what you will) it is in 
fact founded upon the age-frequency distribution of a single 



event in life — its end. It is a notorious fact that Maxwell's 
Demon can reduce all such measures to absurdity, since he can 
strike down perfectly vigorous, or indeed potentially immortal, 
animals at just such ages as will exactly imitate any chosen 
force-of-mortality curve. 

There are many other serious reservations. The use of the 
force of mortality as a measure of senescence assumes that all 



members of the population are equally at risk. This is not true, 
because wage-earners are more exposed to risk than school- 
children or those who have retired. A third difficultv is that if 
a life table is constructed in the way I have suggested — that is 
by following the life histories of a cohort of the newly born — 
it is only too likely to be corrupted by secular changes in the 
hazards of which human beings may be victims. Individuals 
aged seventy to-day were born in 1881, when the causes of 
the death of children, and their likelihood of surviving early 
youth, were very different from what they are to-day. A fourth 



difficulty is that if the population is rather crudely subdivided 
into the innately (that is, genetically) less tough and tougher, 
then the population that reaches age sixty will be by no means 
a genetically fair sample of the cohort with which the life table 
began. Presumably each pattern of genetic constitution endows 
its owners with a characteristic mode of increase of vulner- 
ability; but in a cohort of mixed origins all such distinctions 
must inevitably be confused. 

These are grave difficulties,"* but all of them can be over- 
come in principle, and some in laboratory practice. I now turn 
to a much more important difficulty in the use of vulnerability 
as a measure of senescence; it is ingrained, and in practice 
ineradicable, and it leads us to distinguish between two sorts 
of causes of senescence. 


Consider wrinkles and lines on the skin, for these are familiar 
outward signs of ageing in its colloquial sense. People who often 
frown get lines between the eyebrows; the supercilious reveal 
their temperament by furrows across the forehead; deep lines 
down the corners of the mouth are allegedly the consequence 
of having a ready smile. What is the history of wrinkles? Every 
time one grins or frowns some physical trace is left in the 
cellular or fibrous structure of the skin. These traces are cumu- 
lative, and if only one folds or creases the skin sufficiently often, 
they will add up to form a visible flexure line. One perfectly 
good reason why elderly people should have more lines and 

* [There is another difficulty in accepting 'vulnerability' as a measure 
of senescence: the decline and loss of reproductive power (e.g. in the meno- 
pause) is beyond question a form of senescence, but it is not accompanied 
by any increase of vulnerability in an actuarial sense. I consider this 
problem, and deal more fully with the other difficulties mentioned in the 
text, in The definition and measurement of senescence, Ciba Foundation 
Colloquia on Ageing, Vol. 1, p. 4, 1955.] 



wrinkles is therefore simply that, being older, they have 
frowned and grinned more often. But we must also ask whether 
the skin of older people more readily takes the impress of 
creasing and folding. Does Ql first flexure in the skin of an older 
person leave a bolder trace than a first flexure in the skin of 
someone younger? We may be certain that it does. But the 
point is that both an increase in innate susceptibility to wrink- 
ling and the cumulative effect of recurrent creasing have played 
a part in the history of wrinkles; and although we can dis- 
tinguish the two sorts of causes in theory and in experimental 
practice, they cannot be disentangled merely by contemplating 
the wrinkle as Sifait accompli. 

Wrinkling is an unimportant example of a kind of disability 
that affects all animals. Any injury that leaves a physical trace, 
as all but the most trivial do, increases the vulnerability of 
older animals, because injuries of one sort or another are 
recurrent hazards and older animals, having been exposed to 
them more often, will have built up a bigger actuarial debt. 
Skin scars may be individually trivial things, but the older 
animals will have more of them; and apart from that, germs 
gain easier access to the body during the time taken by a 
wound to heal. Fractures of bone are slow to reunite and 
animals make easy prey until they have done so. The height- 
ened blood-pressure that accompanies the shocks and alarms 
of natural living predisposes the blood-vessels to degenerative 
change. Cells may produce faulty copies of themselves in what 
should be an act of exactly symmetrical division; division is 
recurrent and faulty copies are perpetuated, so that their ill 
effects, summed over the cell population of the body, are bound 
to add up. The efficacy of most of the known cancer-provoking 
chemical compounds depends upon the repeated exposure of 
tissues to their action over long periods. Infections are re- 
current hazards; most infections damage cells, and some do 
permanent damage of a sort that increases vulnerability in an 



obvious way.* To go back to colloquial speech, all these effects 
are the effects of age but not necessarily the effects of ageing; 
they may take their toll even if ageing is not accompanied by 
an innate deterioration. Senescence, as it is measured by 
increase of vulnerability or the likelihood of an individuaPs 
dying, is therefore of at least twofold origin.^ There is (a) the 
innate or ingrained senescence, which is, in a general sense, 
developmental or the effect of '"nature^; and (b) the senescence 
comprised of the accumulated sum of the effects of recurrent 
stress or injury or infection. The latter is environmental in 
origin and thus, paradoxically, the effect of '•nurture\ There is 
always an empirical test for distinguishing between the two in 
principle — one has only to find out whether a, first injury or 
physiological abuse or stress is less well tolerated by older 
animals than their juniors — but in the actual records of vulner- 
ability the two are inextricably combined.^ 

* [Mammals which have what is optimistically described as a 'permanent 
dentition' — i.e. a second and final set of teeth — obviously depend upon its 
remaining in working order; but teeth are bound to get chipped or damaged 
in the ordinary course of biting, and this is a good example of deterioration 
of the kind classified below under heading (6).] 

1 Dr Whitear has pointed out that a third and quite distinct sort of 
change with ageing which influences and will ultimately increase the 
vulnerability of older animals is that entailed by the differential growth and 
changing proportions of the several organs of the body or ingredients of a 
complex tissue. As a general rule, it may be said that every fixed regimen 
of differential growth will, if growth is indeterminate, inevitably lead to 
mechanical or physiological ineptitude of one sort or another, although not 
necessarily involving a loss of 'vitality' at the cellular or tissue level. The 
problem is discussed more fully later. 

2 Higher organisms have means for counteracting the cumulative effect 
of recurrent injuries. Two of the three principal reflex (i.e. response-to- 
stimulus) systems of the body, the immunological and the nervous, have 
the power of 'storing their information' for long periods. The hormone 
system, apparently, has not. In general, an animal is less likely to contract 
a particular infection on its second exposure than on its first, and this is 
mainly due to the fact that what immunologists call the 'secondary' 
response to an immunity-provoking agent is a good deal brisker than the 
first. An animal is also less likely to get bitten, burnt, or otherwise abused 



That one is obliged by the terms of my definition to admit 
that there are two sorts of causes of senescence has, it will turn 
out, no more than a minor nuisance value. I am of course chiefly 
concerned with senescence of sort (a), and you will see that the 
arguments put forward to account for its origin and evolution 
are greatly strengthened by the fact that there may already 
exist a senescence of sort (b). 

The time has now come for a formal definition of senescence, 
and I shall adopt the usual practice of translating a statement 
about the frequency of the occurrence of an event in a popula- 
tion into a statement about the likelihood of its happening to 
an individual. Senescence, then, may be defined as that change 
of the bodily faculties and sensibilities and energies which 
accompanies ageing, and which renders the individual pro- 
gressively more likely to die from accidental causes of random 
incidence. Strictly speaking, the Avord ""accidentar is redundant, 
for all deaths are in some degree accidental. No death is wholly 
'naturaP; no one dies tnerely of the burden of the years. 

By way of an interlude let me now, as a zoologist, apologize 
for appealing so much to evidence from human beings. I do so 
because we know so very much more about the death of human 
beings than of other animals; and though I feel a professional 
obligation to say something about the natural history of 
senescence, there is no time to do so, and even if there were, 
there would not be much to say. 

at each successive exposure to such a hazard; it will have 'remembered' 
the earlier and accordingly learnt better. Two exposures to infection or 
physical risk may therefore have a no more harmful consequence than one, 
and the cumulative effects of some sorts of recurrent stress may therefore 
be to some extent corrected by the benefactions of an immunological or 
nervous memory. Memory, as Professor J. Z. Young has reminded me, is 
also the outcome of some influence that has left a physical 'trace'. 



We can be quite sure that mammals undergo a process of 
Annate'' senescence. But why are we so sure? The answer is 
vital to my later argument. It is because we keep mammals as 
pets, in zoos, and in domestication. If we had to rely upon 
information derived from truly wild animals, we should be very 
much indeed less certain, and it is arguable that we mit^ht 
never know at all. For, as Dr Chitty tells me, wild mammals of 
any perceptible degree of senility turn up in traps so seldom 
that we should always be inclined to think up reasons for their 
enfeeblement that were not necessarily connected with their 
age — the wasting due to infection, maybe, or to an injury that 
stopped them getting food. Animals do not in fact live long 
enough in the wild to disclose the senile changes that can be 
made apparent by their domestication. Many wild birds, as 
Dr Lack has shown, are the victims of so savage an exaction of 
mortality that, beyond a few months of youth, their likelihood 
of dying is actually independent of their age! It is of vital 
importance to remember that senility is in a real and important 
sense an artifact of domestication; that is, something revealed 
and made manifest only by the most unnatural experiment of 
prolonging an animaPs life by sheltering it from the hazards of 
its ordinary existence. Here is a story with a pertinent moral. 
An eminent naturalist was once taken tiger-hunting by a 
courteous Indian potentate; he got his tiger and saw at once 
that it was very, very old. Here then perhaps, he thought, is 
something that he had long vainly looked for — a truly wild 
animal that was very old and very decrepit, and no doubt very 
cunning and very wise as well. On closer inspection, he found 
that the tiger had gold fillings in its molars; the potentate, 
courteous as I said, had simply ''laid it on\ So when you hear 
speak of the 'natural death"* of animals, remember that no 
death is less '■natural"' than that which is commonly so called. 

If there are doubts about mammals and birds, which rep- 
resent the higher classes of vertebrates, how many more must 



there be about the members of what we are now obhged to call 
the under-privileged classes? There is still, it appears, no more 
to be said about senescence in fish than was said by my pre- 
decessor Sir Edwin Ray Lankester some eighty years ago: 
•"Fish are not known to get feeble as they grow old, and many 
are known not to get feebler/ My professional colleagues will 
know that Dr G. P. Bidder held some fascinating and far from 
implausible views on the origin of senescence which turn on the 
belief that fish do not deteriorate with ageing. These I cannot 
delay with. But is it not a most revealing fact that there should 
be any doubt about the matter at all? Fish ma7/ be potentially 
immortal. in the sense that they do not undergo an innate 
deterioration with ageing, and yet the naturalists who ought 
to know about it simply can''t be sure! As you will see, this 
uncertainty is the most tell-tale evidence in favour of my later 
argument. Whether animals can, or cannot, reveal an innate 
deterioration with age is almost literally a domestic problem; 
the fact is that under the exactions of natural life they do not 
do so. They simply do not live that long. 


I have deliberately spent more than half my time in discus- 
sing the measurement and definition of senescence, and I now 
want to discuss the factors that may have played their part in 
its origin and evolution. As a text I shall use a quotation from 
the works of August Weismann. 

Death takes place because a worn-out tissue cannot for ever 
renew itself. Worn-out individuals are not only valueless to the 
species, but they are even harmful, for they take the place of those 
which are sound ... by the operation of natural selection, the life 
of a theoretically immortal individual would be shortened by the 
amount which was useless to the species. 

Weismann''s propositions have the great merit of suggesting, 



for only the second time, that senescence has had a very 
orthodox evolutionary origin. But Weismann is arguing in what 
a student of mine once called a vicious circle, or more exactly 
a vicious figure-of-eight. He assumes that the elders of his race 
are worn out and decrepit — the very state of affairs whose 
origin he purports to be inferring — and then proceeds to argue 
that because these dotard animals are taking the place of the 
sound ones, so therefore the sound ones must by natural selec- 
tion dispossess the old! This is all a great muddle, but there is 
certainly some truth in it, and I shall spend the rest of my 
lecture in an attempt to find out what that truth may be. 

My argument starts with a discussion of certain demo- 
graphic properties of a population of potentially immortal 
individuals, and it will be illustrated by an inorganic model 
which I shall animate step by step. This choice makes it 
possible to avoid two common traps. The first of these is to 
argue that senescence in higher animals has come about because 
they have a post-reproductive period; for ''unfavourable'' here, 
ditary factors that reveal their action only in the post-repro- 
ductive period are exempt from the direct effects of natural 
selection and there is therefore little to stop them establishing 
themselves and gaining ground. Any such argument is wholly 
inadmissible. The existence of a post-reproductive period is one 
of the consequences of senescence; it is not its cause. The 
second trap, into which Weismann fell headlong, is to suppose 
that a population of potentially immortal individuals subject 
to real hazards of mortality consists in high proportion of very 
aged animals with a relatively small number of no doubt brow- 
beaten youngsters running round between their feet. It will 
soon be clear that this idea is equally mistaken. 

I want you now to consider a population of objects, living or 
not, which is at risk — in the sense that its members may be 
killed or broken — but which is potentially immortal in the 
sense that its members do not in any way deteriorate with 



ageing. Test-tubes will do, since they are clearly ""mortar, and 
I shall peremptorily assume that they do not become more 
fragile with increasing age.* 

Imagine now a chemical laboratory equipped on its founda- 
tion with a stock of 1000 test-tubes, and that these are acci- 
dentally and in random manner broken at the rate of 10 per 
cent, per month. Under such an exaction of mortality, a 
monthly decimation, the activities of the laboratory would soon 
be brought to a standstill. We suppose therefore that the 
laboratory steward replaces the broken test-tubes monthly, 
and that the test-tubes newly added are mixed in at random 
with the pre-existing stock. The steward will obviously be 
obliged to buy an average of 100 test-tubes monthly, and I am 
going to assume that he scratches on each test-tube the date 
at which he bought it, so that its age-in-stock on any future 
occasion can be ascertained. 

Now imagine that this regimen of mortality and fertility, 
breakage and replacement, has been in progress for a number 
of years. What will then be the age-distribution of the test-tube 
population; that is, what will be the proportions of the various 
groups into which it may be classified by age? The answer is 
illustrated in Fig. 3. The population Avill have reached the 
stable or ''life-table'' age-distribution in which there are 100 
test-tubes aged 0-1 month, 90 aged 1-2 months, 81 aged 2-3 
months and so on. This pattern of age-distribution is char- 
acteristic of a ''potentially'' immortal population, i.e. one in 

* [In real life, of course, test-tubes could undergo senescence of both the 
types, (a) and (6), which I have distinguished in the text. 'Innate senescence' 
might be represented by the slow crystallization of the glass, which will 
happen whether the test tubes are used or not, and 'traumatic senescence' 
by the accumulation of tiny chips or cracks which, without making the 
test-tube unusable, make it a good deal more likely to be broken in 
everyday use. A life table for glass tumblers has been worked out by G. W. 
Brown and M. M. Flood, Journal of the American Statistical Association, 
42, p. 562, 1947.] 



which the chances of dying do not change with age. The curve 
it outhnes is of a sort very famihar in science. Fig. 3 illustrates 
this very elementary triusm: the older the test-tubes are, the 








monthly aqe groups 
Fig. 3 

fewer there will be of them — not because they become more 
vulnerable with increasing age, but simply because the older 
test-tubes have been exposed more often to the hazard of being 
broken. Do not therefore think of a potentially immortal 
population as being numerically overwhelmed by dotards. 
Young animals outnumber old, and old animals those still 


As a first step in animating this model, I want you to imagine 
that the test-tubes now do for themselves exactly what the 
steward has hitherto been doing for them, i.e. they reproduce 
themselves, no matter how, at an average rate of 10 per cent, 
per month in order to maintain their numbers. Since the popu- 
lation is potentially immortal, the rate of reproduction of its 
members will not vary with their age. It follows that each 
'living' test-tube of the existing population will make the same 



average contribution of offspring to the test-tube population 
of the future. Each test-tube may lay claim to an equal share 
of the ancestry of future generations, and its reproductive 
value is invariant with its age.^ 

The next step in the argument is vital. Although each indi- 
vidual test-tube takes an equal share of the ancestry of the 
future population, each age-group most certainly does not. The 
older the age-group, the smaller is its overall reproductive 
value. The group of test-tubes 2-3 months old, for example, 
makes a very much greater contribution than the group 11-12 
months old. This is not because the test-tubes of the senior 
group are individually less fertile — their fertility is ex hypothesi 
unchanged — but merely because there are fewer of them; and 
there are fewer of them not because they have become more 
fragile — their vulnerability is likewise unaltered — but simply 
because, being older, they have been exposed more often to the 
hazard of being broken. It is simply the old story of the pitcher 
and the well. 

Some of the consequences of this decline in the reproductive 

1 The actuarial characteristics of a 'potentially immortal population' are 

particularly simple: the life table is defined by the relation l^. = /^e"'^^, 

where /^ is the size of the original cohort, l^. is the number of them that 

/' 1 dl^\ 
survive to the age of x, and a is the force of mortality [ {^ = — ) , in- 

\ Ij. dx J 

dependent ex hypothesi of age. The probability Pj. of surviving from birth to 

age X is simply IxJIq ^ 6~'**. If the number of offspring born to each member 

of the population in each unit of age remains constant, as we have supposed, 

at the value h, then the reproductive value remains constant throughout 

1 /'GC h 

life at the value J?~. = — I hpx .dx = — \ and this will also be its value at 

PxJx ^ ^ 

birth (the net reproduction ratio R^. If the regimen of constant mortality 

and fertility has been in progress long enough, and numbers are not 

declining (6> jLt),then a stable age-distribution will be reached in which the 

fraction of the population falling within the age interval a? to a; -H Dx is 

given by Ca; = I he~^^ dx; the proportion of the population aged x 

J X 

and upwards is thus simply e~^^. 



value of older age-groups will be apparent when I take the next 
step in animating my test-tube model. The test-tubes are no 
longer to be thought of as immortal; on the contrary, after a 
certain age, as a result of some intrinsic shortcoming, they 
suddenly fall to pieces. For the time being we shall assume that 
they disintegrate without premonitory deterioration. What 
will be the effect of this genetically provoked disaster upon the 
well-being of the race of test-tubes? It must be my fault if the 
answer does not appear to be a truism — that it depends upon 
the age at which it happens. If disintegration should occur five 
years after birth, its consequences would be virtually neglig- 
ible, for under the regimen which we have envisaged less than 
one in five hundred of the population is lucky enough to live so 
long. Indeed, if we relied upon evidence derived solely from the 
natural population of test-tubes, we should probably never be 
quite certain that it really happened. We could make quite 
certain, as we do with animals, only by domesticating our 
test-tubes, shielding them from the hazards of everyday usage 
by keeping them in a padded box as pets. 

If disintegration should occur one year after birth, an age 
which is reached or exceeded by about one-quarter of the 
population, the situation would be fairly grave but certainly 
not disastrous; after all, by the time test-tubes have reached 
the age of twelve months they have already made the greater 
part of their contribution of offspring to the future population. 
But with disintegration at only one month, the consequences 
would obviously be quite catastrophic. 

This model shows, I hope, how it must be that the force of 
natural selection weakens with increasing age — even in a 
theoretically immortal population, provided only that it is 
exposed to real hazards of mortality. If a genetical disaster that 
amounts to breakage happens late enough in individual life, 
its consequences may be completely unimportant. Even in such 
a crude and unquaHfied form, this dispensation may have a 



real bearing on the origin of innate deterioration with increas- 
ing age. There is a constant feeble pressure to introduce new 
variants of hereditary factors into a natural population, for 
''mutation'', as it is called, is a recurrent process. Very often 
such factors lower the fertility or viability of the organisms in 
which they make their effects apparent; but it is arguable that 
if only they make them apparent late enough, the force of 
selection will be too attenuated to oppose their establishment 
and spread. Such an argument may have a particular bearing 
on, for example, the occurrence of spontaneous tumours and 
the senile degenerative diseases in mice of which Dr Gorer has 
made a special study, for these affections make themselves 
apparent at ages which wild mice seldom, perhaps virtually 
never reach. We only know of their existence through domesti- 
cation; small wonder if they have no effect on the well-being of 
mouse populations in the wild. Mice, of course, do already 
show evidence of deterioration in the course of ageing, but 
my reasoning does not presuppose it. It applies to '"poten- 
tially immortal populations'* with only a quantitative loss of 

It is a corollary of the foregoing argument that the post- 
ponement of the time of overt action of a harmful hereditary 
factor is equivalent to its elimination.^ Indeed, postponement 

^ As an example of what I mean by the time of 'overt action' of genes, I 
should say that the earliest age of overt action of a 'coat colour' gene was 
with the growth of a coat of hair in mice, which are born naked, or with 
birth in animals like the guinea-pig, which are born with a pelt of hair. It 
is not until hairs are both formed and exposed to outward inspection that 
the various coat colours, as such, can influence the welfare of their pos- 
sessors. But I agree with Dr Griineberg that one must be very cautious in 
speaking of the time of action of genes — for one important reason among 
several, because its influence on coat colour may be only one, and by no 
means the most important one, of the manifold actions of what is only for 
convenience of labelling described as a 'coat colour' gene. We have further- 
more only the vaguest ideaof what we mean by speaking of a gene's 'acting' 
at all. This particular difficulty can be overcome by accurate formulation: 



may sometimes be the only way in which eUmination can be 
achieved; but I cannot argue this without an appeal to the 
phenomena of pleiotropy and hnkage, which time Avill not 


It is not good enough to say that what happens to very old 
animals hardly matters and that what happens to youngsters 
matters a great deal. For the degree to which anything may 
matter varies in a predictable way with age, and the selective 
advantage or disadvantage of a hereditary factor is rather 
exactly weighted by the age in life at which it first becomes 
eligible for selection. A relatively small advantage conferred 
early in the life of an individual may outweigh a catastrophic 
disadvantage withheld until later.^ Go back to the test-tube 
model for a moment, and compare two competing test-tube 
populations. Both suffer the same average monthly mortality 
of 10 per cent., and one has, as hitherto, the average monthly 
birth-rate of 10 per cent. The other population has an average 
monthly birth-rate of 11 per cent., but the price paid for this 
hardly profligate increase of fecundity is the spontaneous 
bursting asunder of each member at age two. Which population 
will increase the more rapidly in numbers — that potentially 
immortal, or the mortal population with a birth-rate only 

the time of action of a gene G with respect to a character Cis the age at which, 
in a stated genetic and environmental context, the substitution of G for its 
allelomorph G' transforms the character C into the character C. In short, 
it doesn't matter when (or even whether) G and G' are 'acting' until they 
give evidence of acting in different ways. 

^ By something that is a catastrophic disadvantage to an older animal 
I mean a change which is personally catastrophic, and which would 
certainly be catastrophic to the species as well if it made its appearance in 
younger animals. But in the strict sense, the verdicts 'advantageous' and 
'disadvantageous' can be delivered only after trial by selection, and in this 
sense to speak of 'catastrophic disadvantages' which don't in fact much 
matter is self-contradictory. 



one-tenth part higher than the other's? The simplest calcula- 
tions show that it is the latter. 

A heightened juvenile rate of reproduction, achieved perhaps 
at the expense of recurrent stress that later leads to deteriora- 
tion, is by no means the only possible realization of the 
phenomenon illustrated by this model. It is a general rule, for 
example, that the parts of the body multiply their substance at 
unequal rates, so that proportions change as the body grows. 
There is very likely to be a ""best"* proportion, or a best range of 
proportions, from the standpoint of functional efficiency and 
therefore of survival. In theory these proportions could be 
arrived at once and for all by starting the baby or embryo off 
with the appropriate shape and allowing growth to proceed by 
symmetrical enlargement. This does not happen in practice, 
and it is not biologically feasible for a whole variety of reasons. 
In practice, as I have already said, adult proportions are 
achieved by the adoption of a more or less fixed regimen of 
differential growth, i.e. of a more or less constant ratio between 
the multiplication rates of the several parts of the body. The 
danger inherent in this alternative solution is that there may 
well come a size, and therefore an age, at which proportions 
become functionally and structurally grotesque. The size of the 
male fiddler crab''s claw increases as a power, greater than 
unity, of the size of the rest of its body, and Dr Huxley, who 
has made a special study of these differential growth pheno- 
mena, points out that a crab whose body weighed 1 kg. would 
carry a claw about ten times that weight. But the sense of my 
argument is that if the appropriate proportions are achieved 
at some earlier stage of life, it may not much matter that the 
regimen of differential growth that brought them into being 
should eventually lead to mechanical ineptitude of this degree. 
The early advantage more than makes good the later dis- 
advantage which it necessarily entails. 

E 65 



The postponement of the time of overt action of 'unfavour- 
able** hereditary factors is not just a good idea which the 
organism would be well advised to apply in practice; postpone- 
ment may be enforced by the action of natural selection and 
senescence may accordingly become a self-enhancing process. 
Let me give you a real example in which this process appears 
to be happening at the present time. 

Huntington ""s chorea is a grave and ultimately fatal nervous 
disability distinguished by apparently compulsive and dis- 
ordered movements akin to, and perhaps identifiable with, 
'St Vitus'* Dance'. Its first full clinical description is in George 
Huntington'*s own memoir of 1872, though the evidence I shall 
appeal to comes largely from the fine treatise of Dr Julia Bell. 
Huntington's chorea is a hereditary affliction of a rather special 
sort. Its disabling and clinically important effects first become 
manifest not in youth or old age but at an intermediate 
period, its time of onset — later in men than in women — being 
most commonly in the age-group 35-39. Its age of onset does 
however vary, and I want you to assume (what is almost 
certainly true, though it would be hard to collect the evidence 
for it) that its age of onset, like the disease itself, is also 
genetically determined. 

If differences in its age of onset are indeed genetically deter- 
mined, then natural selection must so act as to postpone it: for 
those in whom the age of onset is relatively late will, on the 
average, have had a larger number of children than those 
afflicted by it relatively early, and so will have propagated 
more widely whatever hereditary factors are responsible for the 
delay. But as the age of onset approaches the end of the 
reproductive period, so the direct action of selection in post- 
poning it will necessarily fade away. 

One may now ask why, if such a thing must happen, has it 



not happened already, and, if it has not, what is the evidence 
that it is happening now? The first question amounts to asking 
why Huntington's chorea is not already one of the diseases of 
the post-reproductive period, since selection of the sort I have 
outlined must be pretty vigorous and has presumably had tens 
of thousands of years at its disposal. My answer to this is based 
on an aside of Professor Haldane''s. It is only in the last century 
or so that selection has had a real chance to get a grip on it, 
for it is only within this period that the average expectation of 
life at birth has come to equal the average age of onset of the 
disease.* Even so, there is indirect evidence of a postponement 
of its age of onset. Since the male reproductive span is longer 
than the female''s, the force of selection on men must be less 
quickly attenuated with increasing age; postponement should 
therefore have gone farther in men than in women — and this, as 
I have already said, is indeed the case. Ultimately, no doubt, the 
age of onset will come to a standstill in both men and women 
at the end of their respective reproductive periods. I gratefully 
acknowledge the origin of this train of thought in Professor 
Penrose''s writings on mental disease and natural selection. 

With Huntington''s chorea as a lucky concrete example, I 
can now propound the following general theorem. If hereditary 
factors achieve their overt expression at some intermediate age 
of life; if the age of overt expression is variable; and if these 
variations are themselves inheritable; then natural selection 
will so act as to enforce the postponement of the age of the 
expression of those factors that are unfavourable, and, corre- 
spondingly, to expedite the effects of those that are favourable 
— a recession and a precession, respectively, of the variable 
age-effects of genes. This is what I mean by saying that 

* [This is not quite fair. It is not the mean expectation of life at birth 
that is important, but the mean expectation of further life at an age when 
reproduction has just begun. This too has increased, but not nearly so 
dramatically, over the past hundred years.] 



senescence is a self-enhancing process. The theorem in the form 
in which I have just put it does not depend upon the existence 
of a post-reproductive period; it only requires that the repro- 
ductive value of each age-group should diminish with increas- 
ing age. I have argued that this must necessarily diminish even 
with a population of potentially immortal and indeterminately 
fertile individuals, provided only that they are subject to real 
dangers of mortality. In such a population a younger age-group 
must necessarily outnumber an older, for the older represents 
the residue of those who have been longer exposed to mortal 
hazards. If you should have, as I believe, unjustified qualms 
about an argument based upon combining an innate potential 
immortality with a contingent real mortality, I would recall to 
you my earlier distinction between senescence of sorts (a) and 
(b). Senescence of sort (b) is not innate or 'laid on'' develop- 
mentally; it represents the outcome of the cumulative effects of 
recurrent physical damage, physiological stress, or faulty 
cellular replication. If you will admit that senescence of this 
sort is a means by which, irrespective of any genetical back- 
ground, the reproductive value of each individual in a popula- 
tion is caused to diminish with increasing age, then my argument 
is quantitatively strengthened, because the numerical pre- 
ponderance of the younger age-groups will become so much the 
more pronounced. And if, further, a post-reproductive period 
of life is already established, then indeed it becomes, as it were, 
a dustbin for the effects of deleterious genes. But these pro- 
positions are mere glosses or refinements. The argument must 
stand or fall on the case which I first proposed. 

I have now suggested three agencies which may have played 
a part in the evolution of ''innate'' senescence: (1) the inability 
of natural selection to counteract the feeble pressure of 



repetitive mutation when the mutant genes make their effects 
apparent at ages which the great majority of the members of 
a population do not actually reach; (2) the fact that the 
postponement of the time of action of a deleterious gene is 
equivalent to its elimination, and may sometimes be the only 
way in which elimination can be achieved; and (3) the fact 
that natural selection may actually enforce such a postpone- 
ment, and, conversely, expedite the age of onset of the overt 
action of favourable genes. All these theorems derive from the 
hypothesis that the efficacy of natural selection deteriorates 
with increasing age. 

I am inclined to think that the third factor, the enforced 
precession and recession of the ages of the overt action of genes, 
has the widest ambit of significance. But although I have fore- 
sworn the introduction of too many qualifying and saving 
clauses, one indeed is most important. Real animals, unlike 
imaginary test-tubes, are neither born mature, nor do they 
get on with the business of self-reproduction at once. There is 
always a pre-reproductive period during which animals are far 
from exempt from the hazards of mortality, and during this 
period the average reproductive value of an individual must 
therefore rise to a maximum, irrespective of whether or not it 
falls later. If my reasoning is correct — there is no time to go 
into details — the precession of the time of action of genes 
comes to a standstill at the epoch when the reproductive value 
is at a maximum, and it is then that senescence should be 
expected to begin. Professor R. A. Fisher has pointed out that 
the actuarial prime of life of human beings and the age at 
which their reproductive value is at its maximum do in fact 
nearly coincide. 

Even with such refinements as this, my proposals can hardly 
be said to add up to a self-sufficient theory. If we concede that 
the force of natural selection is rather exactly weighted by the 
ages of the animals on which it operates, it is still far from easy 



to see in detail how senescence has become shaped into its 
distinctive pattern — the early onset and slow progressive fulfil- 
ment that the curve of the force of mortality so conspicuously 
reveals. Some of the agencies described seem to suggest a 
rather precipitous onset of senescence — more like that which 
befell the expatriates of Shangri-La than that suffered by the 
inhabitants of the world at large. But even allowing this short- 
coming, I think it must be clear that the origin and evolution 
of senescence is not an insoluble genetical mystery, however 
mysterious it may be in other ways. The geneticist can see how 
it might well have happened; its occurrence does not outrage 
his sense of the fitness of things. So perhaps I was unduly 
disrespectful to Weismann's memory when I poked fun at his 
conjectures on senescence. In very broad outline they were 
probably not erroneous, at least in so far as natural selection 
was recognized as the instrument of its origin and perpetuation. 
I said earlier, as you may remember, that there was some truth 
amidst a good deal of what we can now see to be nonsense, and 
that it would stir up his successors to think up a more polished 
and cogent explanation. Not much more than this can be said 
of any biological theory of comparable pretensions, and I shall 
count myself lucky if I hear an equally sympathetic criticism 
of my own. 


A Note on 
' The Scientific Method ' 

It now seems to be agreed by those who direct our poKcy that 
the development and application of science is of immediate 
importance to England''s economic welfare. So long as science 
could be thought of only as a means for the leisurely inception 
of an Age of Plenty, its benefactions could be postponed with- 
out fatal consequences. But we must now be satisfied with 
lowlier aspirations: science is to lead the state as the Red 
Queen led Alice — the most rapid progress is necessary with no 
higher ambition than to remain in approximately the same 
place as before. We now, therefore, hear a great deal about *"the 
scientific method**, for the most part from people who might be 
quite upset if they were asked just what that method was 
supposed to be. The scholarly amateur might be heard to 
mumble something about the Question put to Nature and the 
experimentum crucis; the scientist speaks of quantitative method 
and the controlled experiment; the layman is often rude 
enough to think it no more than common sense. Let us 
press the question. How does scientific method differ from 
that used in other sorts of scholarly enquiry? What are the 
rules for making scientific theories? Just what does science 
prove? The answers to these questions have been quite 
widely agreed upon, but are not yet common property; they 
should be, and this essay is an attempt to make them so. 
Being no philosopher myself, it goes without saying that in 



what follows I claim proprietary rights only in what may be 

The currency of science consists of statements about 'matter 
of fact and existence** — propositions, they are often called, to 
distinguish them from questions, orders, outcries and sugges- 
tions, and some forms of the expression of abuse. But scientific 
knowledge is something more than the assembly of the facts 
reported by such statements: it has a corporate structure, a 
certain internal order and coherence of its own. There are 
several ways in which an order might be imposed upon them. 
For example, a man who wished to write a textbook about 
boron might begin by collecting under that heading all true 
statements made about it. This would not give the facts a 
peculiarly scientific structure, because the man who wrote the 
author*'s obituary notice would be expected, mutatis mutandis, 
to do very much the same for him. The grouping of statements 
by their subjects, objects, form or syntax, or the chronological 
order of the events recorded in them, though each has its 
special purpose, does not confer the structure of a theory upon 
them. Theories are sets of statements put into order by the 
relationship of entailing, and statements entailed are said to be 
•"explained' by those they follow from. Statements at the head 
end of the entailing are variously called premisses, axioms, 
postulates, or hypotheses — a luxuriant s^Tionymy, since all are 
in effect, though not in form of origin, the same. Some attempt 
should be made to share out their legacy of meaning without 
spreading dissatisfaction equally among them all. We assert a 
postulate, and take an axiom for granted; hypotheses we merely 
venture to suggest. 'Premisses'*, when other people''s, are usually 
so spoken of when not believed in. Scientists speak as a rule of 
their hypotheses. Some scientists seem to use the word 'theory"* in- 
terchangeably with 'hypothesis'*, but this wastes a good word and 
should not be encouraged. A theory is the whole system of state- 
ments comprising hypotheses and the statements they entail. 



All this is commonplace and rather uninspiring. What is of 
intense personal interest to many scientists is how an hypothesis 
ever comes to be devised at all. Its creation is evidently a leap 
upstream of the flow of deductive inference. One does not, as 
writers of detective stories seem to imagine, deduce hypotheses; 
quite the reverse, hypotheses are what we deduce things from. 
It was at one time thought that hypotheses could be arrived at 
by a rigorous logical process of ''induction'', but even that 
humblest sort of hypothesis (for such it is), the simple collective 
generalization, defied these efforts to make it logically respect- 
able, and it defies them no less resolutely to-day. Philosophers 
who now irritably contend that induction does not require 
their formal blessing forget that it was they themselves or their 
predecessors who first attempted the laying on of hands. Leav- 
ing aside those forms of scientific enquiry that may be purely 
documentary or descriptive in purpose — the determination of 
an atomic weight, say, or the anatomy of a mollusc — it seems 
that no attempt to solve a scientific problem can even be begun 
without the subsidy of some hypothesis, however dimly formu- 
lated or however vague. The first stage of textbook induction 
as I learnt it used to be the assembly of 'relevant"* information; 
but what could it be relevant to, if not to the terms of some 
preconceived hypothesis? In my experience there is no stage in 
the working out of a scientific problem in which some hypo- 
thesis is not for the time being in office, and scientific activity 
comes instantly to a standstill without this sort of direction of 
its affairs. This, of course, says nothing about how hypotheses 
come into being. So far as I can tell from my own experience 
and from discussion with my colleagues, hypotheses are thought 
up and not thought out. One simply ''has an idea** and has it 
whole and suddenly, without a period of gestation in the 
conscious mind. The creation an of hypothesis is akin to, and 
just as obscure in origin as, any other creative act of mind. If 
science were an art we should call it inspiration, but as only 



astronomy has a Muse that will not do. Our leading phil- 
osopher^ once called it 'a mere method of making plausible 
guesses\ The word mere rankles, for it is guesswork that must 
be imaginative, apt and technically informed. Too much learn- 
ing may however be as dangerous as too little. Ail scientists 
know of colleagues whose minds are so well equipped with the 
means of refutation that no new idea has the temerity to seek 
admittance. Their contribution to science is accordingly very 

It is right to point out, because of the irritating mystique that 
has grown up round it, that clinical diagnosis illustrates the 
act of hypothesis formation in an uncomplicated and fairly 
lucid way. The clinician seeks an hypothesis that will account 
for his patient's illness. There is no time in the course of his 
investigation during which some hypothesis is not in the back- 
ground of his mind, and during its early stages there may be 
many. If his mind ends up blank after examination, that is not 
because no hypothesis sought admittance, but because all that 
did so had to be turned away. The experienced clinician is very 
well aware of the intuitive nature of the act of mind by which 
he hits on an hypothesis, but he sometimes fails to realize 
that this is the commonplace of scientific discovery: hence the 

A scientific theory is propped up on either side, like Moses"* 
arms before the Amalekites, by twin supports that together 
form its ""metatheory"*, and without these Reason cannot pre- 
vail. One part of metatheory, now called logical syntax, deals 
with the concepts of formal truth and falsity and the ordin- 
ances that govern the activity of deducing. Logical syntax is 
wholly the logician's business. The second part, semantics, 
more recent of origin and in lay circles now more fashionable, 
deals with the theory of the meanings of words and the ideas of 
material truth and falsity. The semantic problems of a science 
1 Bertrand Russell: The Principles of Mathematics, p. lln, 1903. 



have always been solved and are best solved by its own prac- 
titioners, and no more need be said about them here. But 
several things are Avorth saying about deduction. From the 
days of Sextus Empiricus onwards philosophers have con- 
fidently or more or less reluctantly affirmed that the process of 
deduction is simply the unravelling of tautology. Deduction 
renders explicit, discloses or makes manifest the information 
concealed within the axioms from which it issues; so far from 
adding new information, it merely attenuates it or makes it 
more dilute.* Thus the theorems of Euclid are but a few of the 
endless possible reaffirmations of his axioms; they exist as 
reproachful evidence of the mind"'s imperfection, because for 
a perfect mind the axioms would be enough. Deduction in- 
volves no creative act of mind and no imagination. The Mech- 
anical Brain will one day undertake our deductive reasoning 
for us; to some extent it already does. The respect that our now 
queasy Frankensteins show for their intricate but guileless 
monster may be due to their realization that mathematics is 
but tautology after all. If that is so, it will serve them right for 
the qualms they have caused among the laity if such a Brain 
one day submits its candidature for the Wayneflete or Sadleir- 
ian Chair. 

A second property of deduction is of the utmost importance 
for appreciating the validity of what is so often recklessly 
spoken of as ''proof\ The rigours of deduction are in one way 
curiously overrated: it proves to be quite a lenient discipline 
after all. For when it is said that one state ment^/bZZoz£'*yrom 
another, deduction admits any combination between the truth 
or falsity of either except just one: that the first statement 
should be true and the second false. All that it guarantees 
among alternative possibilities is that what follows from a true 
premiss should be true. The dilemma of ""proof is simply that an 

*[H. A. Rowlands has put it admirably: deduction obeys a Law of 
Conservation of Knowledge.] 



hypothesis may be false although the inferences drawn from it 
are themselves empirically true. This combination is by no 
means disalloAved by logic. Consider that paradigm of empirical 
facts, the mortality of Socrates. If Socrates is a fish and all fish 
mortal, it follows with pitiless logical rigour that Socrates is 
mortal too. This is perfectly straightforward deduction, and 
Socrates'* mortality is adequately so explained. How then does 
the scientist prove his hypotheses? The answer, now widely 
accepted, is that except in certain limiting instances he never 
does. No concept is so maltreated by lay usage as that of proof. 
In a strictly formal sense, accepted hypotheses remain per- 
petually on probation; one does not prove them true, though 
one may often act exactly as if they were. But what the scientist 
can often do with complete logical precision is to disprove 
hypotheses. If what follows from an hypothesis is false, then the 
hypothesis is false, and false in logic. This consequence of the 
asymmetry of the process of Smplying** is a central property of 
scientific method, and it influences experimental design in a 
direct and conspicuous way: many experimental designs are 
simply well-laid traps to lure on a so-called 7iull hypothesis and 
then confound it. The precision of the act of disproof is thus 
very far from being a formalistic fancy. This does not mean, of 
course, that the accepted hypothesis is merely 'not disproven"*; 
there are obviously degrees of certitude of conviction, but these 
are for the most part informally worked out. It is clear that 
an hypothesis gains in acceptability merely by its fitting in to 
a wider theoretical scheme of which it is a part. In this way 
hypotheses may sustain each other.* At all events, the scientist 
would soon be beggared by Descartes'* first precept of intel- 
lectual enquiry- -""c/e ne recevoir jamais aucune chose pour vraie 
que je ne la con?iusse evidemment etre telle''. His own precept of 

* [I confess here to the fault of having lumped all hypotheses together, 
as if they were a single logical species; for a careful analysis of their several 
forms, see Probability and Induction, by W. Kneale, Oxford, 1949.] 



enquiry is the mirror image of this one, to accept nothing 
which is demonstrably false. 

No hypothesis is admissible in science that accounts only for 
the facts it was expressly formulated to explain. Such an hypo- 
thesis is inadmissible not, as we have seen, because it cannot be 
verified but because it cannot even in principle be proved 
untrue. Nothing can be done with an hypothesis that has no 
'extra-muraP implications, and its acceptance and rejection 
are equally acts of faith. 

'Testing an hypothesis' is the act of examining these extra- 
mural implications. If they are true, the hypothesis is in some 
recognizable but obscure way strengthened; if they are false, 
the hypothesis is false. All fish have gills, but Socrates proved 
to lack them; we must therefore think up some other explana- 
tion of his having died. In practice, of course, we are not often 
lucky enough to deal with such crisp disjunctions. An hypo- 
thesis is less often outright false than merely inadequate, and 
not beyond the help of running repairs. 

The act performed to test an hypothesis may be called an 
*'experiment\ It is best to use the term in this simple and clean- 
cut way, rather than to follow common use in restricting its 
terms of reference to some sort of active messing-about with 
nature. A ""mere observation** may in this sense be an experi- 
ment, and if activity is insisted upon as a criterion, it may be 
answered that even the merest observations cannot be made 
from a supine position. The hypothesis which predicted the 
existence of a planet Neptune was tested by the experiment of 
directing a telescope towards a certain predicted region of the 
sky. The existence of an experimental method in this generalized 
sense is what distinguishes the scientific method from that of 
any other sort of scholarly enquiry, and it is to this method that 
science owes its power. It will be noticed that I have said 
nothing here about the virtues of metrical analysis or the con- 
trolled experiment, or all the many other things that to the 



layman seem to be so characteristically scientific. These things 
belong not to scientific method in its more formal sense but to 
the theory of experimental design and scientific analysis, and 
the exacting requirements of scientific reportage. Obviously an 
experiment must be done in such a way as to give an un- 
ambiguous answer, and in the examination of events one must 
aspire to put on record that which is indeed the case. All this 
belongs to the technology of scientific method. 

The growth of science is organic and not accretionary. The 
structure of knowledge built up by the prosecution of the 
scientific method as I have outlined it is a tapering hierarchy 
of hypotheses, the more general counting the less general 
among their consequences, the least general — the ordinary 
colligative inductions — finally touching down in a multitude of 
particular statements about fact. The structure of scientific 
knowledge is therefore in the outcome logico-deductive, and 
this is the form in which what Berkeley called the Grammar of 
Nature is finally wTitten down. It is already a record of some 
grandeur, though the greater part has yet to be compiled. 


A Commentary on Lamarckism 


I begin by excusing myself the task of making any detailed 
exposition of the evolutionary teachings of the Chevalier de 
Lamarck. Darwinism, we know, is Wallace''s word, and Lam- 
arckism is not Lamarck*'s; and although the Avords stand for 
doctrines which their eponymous authors would have no diffi- 
culty in recognizing as their brain-children, it is their latter-day 
growth and present stature that must occupy the whole of our 
attention. Nor will it be profitable to carry out a semantical 
autopsy upon expressions like *'the survival of the fittest' and 
'the inheritance of acquired characters\ "'Fitness'' is now so 
defined as to make Spencer''s phrase a tautology, and its use 
tends to perpetuate the mistaken belief that the famous Mal- 
thusian syllogism is a necessary part of the logical structure of 
Darwinism (see Fisher, 1930, p. 43). As to the 'inheritance of 
acquired characters', its last solemn rites have been capably 
intoned by Woodger (1952), and there can be no case for having 
it disinterred. 

The purpose of this introductory section is {a) to present in 
the simplest possible terms the essential difi'erence between 
Darwinian and Lamarckian interpretations of the hereditary 
process, and (6) to show that inheritance that may be repre- 
sented as Darwinian on one plane of analysis may be repre- 
sented as Lamarckian on another. 

Consider for this purpose a population of streptococci 



(though other micro-organisms will also serve). Streptococcal 
infections are usually treated by the administration of sulphon- 
amide drugs or of antibiotics of fungal or bacterial origin, such 
as penicillin or streptomycin. It is a common observation of 
clinical practice (and one which can be reproduced by experi- 
ments in vitro) that the prolonged exposure of a population of 
streptococci to penicillin, at concentrations which fall short 
of bringing about its complete destruction, mai/ lead to the 
evolution of a ''resistant strain\ i.e. a population which can 
flourish unchecked at a concentration of penicillin that strongly 
inhibited the growth of the parental organisms. The trans- 
formation is heritable, for resistance once acquired long out- 
lives the stimulus that originally called it forth. 

By disregarding all subtleties of interpretation, the ''training'' 
process may be represented in alternative ways. The first is 
illustrated by Fig. 4; the shaded circles represent resistant 







O Q 

Fig. 4. The development of penicillin-resistance in 
bacteria, according to a Darwinian interpretation. 
For explanation see text. 

forms. It is presumed that the original bacterial population 
was heterogeneous and contained genetic variants endowed 
with a relatively high degree of resistance to the action of 
penicillin. In the normal course of events — in a penicillin-free 



microcosm — these variants stand at no selective advantage; 
but under the influence of penicilHn they proliferate more 
rapidly than their unresistant neighbours and so eventually 
become the prevailing forms. 

The second is illustrated by Fig. 5. It is presumed that the 
enzymic organization responsible for the metabolic activity of 







Fig. 5. The development of penicillin-resistance in 
bacteria, according to a Lamarckian interpretation. 
For expanation see text. 

each individual is progressively altered, as the shading indi- 
cates, and that the change so produced is heritable. The change 
in the properties of the population is therefore the sum of the 
changes brought about within each individual. The difference 
between the two interpretations, Darwinian and Lamarckian, 
is that the one presents adaptation as a change in the genetical 
structure of a population, and the other as a change in the 
genetical structure of an individual. These are avowedly ex- 
tremes, for they are in no sense mutually exclusive. On the 
contrary, any inherited difference in the 'Lamarckian'' adapt 
ability of individuals must of necessity become the subject of 
selective discrimination. 

So much is commonplace. Let us now consider not a single 
population but an assembly of such populations, supposing 
each one to be anatomically separate and distinct. The entire 

F 81 


assembly of populations is now subjected to training by peni- 
cillin, and it is found that each individual member becomes 
progressively adapted to resist its action. In a world in which 
such populations were the analytical units, such a transforma- 
tion would be called ''Lamarckian'' in whatever sense the scheme 
illustrated by Fig. 5 may be so described. But within each 
population, the adaptive change might very well be of the type 
illustrated in its simplest form by Fig. 4. 

This reflection is instructive if we return to consider the 
activities that may be supposed to accompany the transforma- 
tion of an individual bacterial cell. It may be assumed that 
there are alternative pathways of metabolism within each cell, 
i.e. alternative enzyme sequences or metabolic gearings, as 
there are, for example, alternative pathways for the degrada- 
tion of glucose. Such metabolic pathways may for a variety of 
reasons be so adjusted as to be mutually inhibitory, so that 
only one prevails in any one of a possible set of steady states. 
The inhibition of one such system therefore entails its replace- 
ment by another. In other words, as Hinshelwood (1946) has 
made clear, the Lamarckian transformation illustrated by 
Fig. 5 may be Darwinian at the lower analytical level repre- 
sented by the enzymic population or complex of intersecting 
metabolic pathways wuthin the individual bacterial cell. Such 
a description would be pointless for any except explanatory 
purposes, but it shows that no discussion of the rival inter- 
pretative powers of Darwinism and Lamarckism can have any 
useful outcome unless a certain analytical level is defined and 
adhered to. Hereafter we shall be concerned with individual 
organisms as analytical units, for it is only in this context that 
the rivalry is of any moment. 

The case for and against Lamarckism may be set out for 
analysis in a variety of ways. Guided by the reflections of 
Baldwin and Lloyd Morgan, I shall present it first in what 
philosophers would call a '"weak"' or general form, and then in a 



'strong"* or special form. The weak form may be so described 
because it merely proposes the existence of a certain mode of 
origin of inherited differences, without expressing any opinion 
about the actual mechanism by which those diiferences have 
come into being; but they are diiferences which, unlike so 
many, are open to a Lamarckian interpretation of their origin. 
The 'strong'' form goes farther and positively affirms that the 
Lamarckian interpretation left open by the weak formulation 
is in fact the correct one. 


The 'weak** form, then, may be expressed in these terms: 

Modifications acquired in each member of a succession of indi- 
vidual lifetimes, as a result of recurrent responses to environmental 
stimuli, may eventually make their appearance in ontogeny even 
when the environmental stimuli are absent or are deliberately 
withheld. . . . 

We may proceed at once to strengthen this formulation by 
making it in one respect a little more particular: 

. . . and the age of appearance of these modifications in ontogeny 
will eventually anticipate the age at which environmental stimuli 
could in any case have been responsible for them 

This clause is separated from the main body of the formula- 
tion merely to emphasize the fact that it is formally separable, 
but we shall adopt the fuller and more particular formulation 
for the good reason that every example we shall consider will be 
shown to satisfy it. It must again be emphasized that the 'weak** 
formulation neither embodies nor presupposes any hypothesis 
about how acquired character differences become inherited 
character differences: it merely states that they do in fact 
become so. 

Before proceeding to the discussion of special examples, we 
may ask: of which character differences may it plausibly be 



argued that they have arisen in the way that has just been 
proposed? The answer is a very simple and obvious one: they 
are character differences having the distinctive property that, 
although they are in fact ''laid on** by development, they could 
in any event have been fashioned in a?i individuaVs own lifetime 
merely as a response to differences of use. 

Consider, for example, the difference between the character- 
istically thick, richly stratified and mitotically active epidermis 
on the sole or heel of the foot and the thinner and more delicate 
epidermis that covers the greater part of the rest of the body. 
The difference is at least in large part of purely developmental 
origin, i.e. of the same sort as that which distinguishes epi- 
dermal cells from pancreatic or thyroid cells. It does not arise, 
as we are at first tempted to think, because of the chronic 
chafing and general mechanical stress that soles of feet are 
obliged to put up with (although such stimuli can certainly 
exaggerate the difference). Both the human being and the 
guinea-pig are born with a thicker epidermis on the sole of the 
foot than elsewhere on the body.* Such a difference is therefore 
developmentally prefabricated; it could not have arisen as an 
adaptive response in utero because the foetus treads water in 
so far as it treads at all. 

The argument may be reinforced by experimental proof. If 
the difference between trunk and sole-of-foot epidermis arose 
merely because the latter is habitually trodden upon and other- 
wise abused, while the former is not, then sole-of-foot epidermis 
should revert to the condition of relatively delicate and 
quiescent body skin after transplantation to a protected posi- 
tion elsewhere on the body. Billingham and I (1948, a, ^), have 
done this experiment on the guinea-pig, and find that sole-of- 
foot skin conserves its distinctive thickness, stratification and 
mitotic activity even two years after its transplantation to a 

* [A fact well known to Darwin, and commented upon by others since: see 
C. H. Waddington's The Evolution of Adaptations in Endeavour, 12, 1953.] 



completely protected position on the ordinary skin of the chest. 

So much for the evidence that the difference between sole 
and body epithelium is of developmental origin, i.e. is an in- 
herited difference between the somatic cells that arise by 
fission of the zygote. It must now be shown that even if the 
difference were not of developmental origin, it would be almost 
exactly reproduced within an individuaPs own lifetime as a 
response to differences in the habit of use; it must be sho^vn 
that if guinea-pigs or human beings were in fact born with a 
thin and delicate epithelium on the soles of their feet, ordinary 
use would soon toughen and thicken it. 

There can be no reasonable doubt that this would be so, 
because a normally quiescent epidermal epithelium can easily 
be induced to thicken in response to chronic irritation. Corns 
are so formed on the thin skin of the dorsum of the toes; 
callosities on the hands develop as a response to chronic chafing. 
A com has a histological structure very closely similar 
to that of the skin on the heel of the foot, with a deep, strati 
fied, vigorously dividing epidermis, a thick pad of compact 
cuticle, and tall, steeply rising dermal papillae. The difference 
is that corns and callosities do not last much longer than the 
mechanical stimuli that provoked their formation; corns sub- 
side with the wearing of shoes that fit; callosities may be cured, 
as they may also be avoided, by wearing gloves. Evidently the 
epidermis has the capacity to thicken in response to mech- 
anical abuse. In the epidermis of the sole or heel, this thicken- 
ing is developmentally anticipated and does not depend for its 
maintenance upon the continued stress of use; and yet, were 
it not so anticipated, stress of use could be relied upon to 
reproduce it faithfully. All adaptations that are open to a 
Lamarckian interpretation have this distinctive character: 
that they represent differences of developmental origin that 
can be faithfully mimicked within an individuaPs own lifetime 
by differences in mode of use. 



Contrast the state of affairs that has just been described 
with another difference between the races that together con- 
stitute the epidermal (or ectodermal) epithelia: the difference 
between '"ordinary'' body skin and the compact, non-flaking, 
perfectly transparent epithelium of the cornea. Here again, 
the difference is of developmental origin; nor is it kept in being 
by the fact that the corneal epithelium lives in an environment 
very different from that of ordinary skin. The cornea is non- 
vascular, moist and cool; ordinary body skin is vascular, dry 
and (being dry) warmer than the cornea. Yet if corneal epi- 
thelium is transplanted to an area formerly occupied by 
ordinary body skin, and vice versa, the distinctive differences 
between the two remain (Billingham and Medawar, 1950). The 
property that distinguishes this case from the one just con- 
sidered is this: that if the difference between corneal and ordin- 
ary body-skin epithelium were not of developmental origin, 
it could 7iot be reproduced within an individuaPs own lifetime 
by difference of environment or of mode of use. The difference 
will be established by developmental mechanisms, i.e. by the 
appropriate segregations within the lineage of cells arising by 
division of the zygote, or not at all. 

Let us call the difference between corneal and body-skin 
epithelium a difference of Class A, and that between sole-of- 
foot and body-skin epithelium a difference of Class B. To these 
should be added a third category of difference, of Class C 
(Abercrombie, 1952): one which is not developmentally pre- 
fabricated, but which may arise purely from difference of 
environment or of use.* The pigmentary cells (melanocytes) of 
the epidermis of the two sides of the face or the two arms may 

* [This rather arid terminology was based upon that of an article 
published in New Biology, 11, p. 10, 1951. Much better, because self- 
explanatory, is C. H. Waddington's {loc. cit., 1953): Class C adaptations 
are 'exogenous'. Class A 'endogenous', and Class B 'pseudo-exogenous'. 
I was not able to benefit from these suggestions, because the present article 
was two years 'in the press'.] 



be supposed to have the same properties and to be present in 
the same numbers. Expose one side of the face or one arm to 
sunhght, and it will become darker than the other, no matter 
why. The difference of degree of pigmentation is caused by 
differences of environmental stimuli, and by them alone. 

Differences of Class B, those which are open to a Lamarckian 
interpretation of their origin, are commoner than is usually 
supposed. The flexure lines of the palm of the hand provide a 
splendid example. The bolder flexure lines are easily visible in 
the twelve weeks' foetus, and even if the foetus is not incapable 
of clenching its hands, it would be idle to suppose that flexure 
lines were formed by the imprint of habitual use. Use neither 
forms them nor keeps them in being, for the plastic surgeon 
tells us that if skin grooved by a flexure line is displaced or 
transplanted to positions in which it is not normally creased 
or folded, the flexure lines will nevertheless persist. Yet ectopic 
flexure lines can be formed by habitual creasing of the skin — 
by frowning for example, or raising the eyebrows; and we can 
therefore be quite confident in saying that if the palmar flexure 
lines were not developmentally prefabricated, a very exact 
copy of them would soon be formed in the ordinary run of 
everyday use. Ectopic flexure lines are exactly analogous 
to the corns and callosities that were called in evidence in our 
earlier example. They differ from the Suborn" flexure lines 
because they disappear with the withdrawal of the stimulus 
that was responsible for their formation. 

In saying that flexure lines and thickened soles are of 
developmental origin, I do not wish to deny that use within 
an individual's own lifetime may not make flexure lines bolder 
and soles thicker still. Why should it not be so, if folding and 
chafing of the skin can cause the formation of ectopic flexure 
lines or epidermal thickenings elsewhere on the body? It is 
likely, but not certain, that the ordinary use of a joint or bone 
completes the otherwise purely developmental differentiation 



of articular surfaces and the patterns of bony trabeculae. With 
this quahfication, the mode of development of joints is closely 
comparable to that of flexure lines and thickened soles. We are 
born with working joints, cartilage lined, encapsulated, lubri- 
cated with synovial fluid, and with their apposed surfaces 
having just that complementarity of structure which might be 
expected to arise from the mechanical exactions of ordinary 
use. Foetal movements have only a small part to play in 
fashioning the final structure, for joints develop from primor- 
dia cultivated ^7l vitro or transplanted to the chorio-allantoic 
membrane — positions where no movement can occur. In spite 
of that, functional cartilage-lined and encapsulated ectopic 
joints can be formed in an individuaPs later lifetime if by 
accident (or orthopaedic artifice) two mobile bony surfaces are 
apposed to each other, as in an unhealed fracture (see Le Gros 
Clark, 1952). Here too then, it appears, the mechanisms of 
morphogenesis exist in duplicate, and what could be formed 
by use is in fact formed by pre-emptive diff'erentiation. 

In the foregoing account I have deliberately confined myself 
to familiar everyday examples of pre-emptive diff'erentiation in 
metazoa. (Micro-organisms come later.) The more esoteric 
examples collated by Wood Jones (1943) ^ appear to me to 
introduce no distinction of principle, and an explanation valid 
for the one set should be valid for the other. Each represents a 
character diff'erence of developmental origin that could also 
have arisen as a direct adaptive response to difference of use 
within an individuaPs own lifetime. 

All such adaptations are open to a Lamarckian interpreta- 
tion of their origin. All that remains to establish a strong /?m«a 
facie case is evidence that acquired character diff'erences can 

1 For example the squatting facets between tibia and ankle-bone in 
Panjabi (but see Medawar, 1952) and the callosities on the 'knees' of the 
African wart-hog; to which add Kukenthal's strange story of the dugong's 
teeth, as it has been recounted by de Beer (1951). 



become inherited character differences under conditions that 
formally exclude the action of natural selection. It will be clear 
from Section 3 that evidence of the occurrence of any such 
transformation in metazoa is still wanting. We must conclude 
that although what I have called 'Class B' adaptations might, 
unlike so many others, have arisen in Lamarckian fashion, 
there is no unambiguous evidence that they have done so. 

This answer is quite widely thought by laymen and ill- 
informed zoologists to be shifty-eyed and evasive, and the 
reason is not far to seek. It is believed, quite mistakenly, that 
eligibility for a Lamarckian interpretation is in some way dis- 
creditable to Darwinism. The truth is quite otherwise. The 
developmental pre-emption of what would otherwise be ac- 
quired character differences is, with most adaptations of Class 
B, of conspicuous selective advantage. If it is an advantage to 
have thickened soles at all, it will be particularly advantageous 
to have them ready-made — ready for use the first time the foot 
touches the ground. And what could be more biologically inept 
than a state of affairs in which the several joints, only roughly 
fashioned at birth, had to be ''run in'' to complete their par- 
ticular articulation patterns during the lifetime of each 
individual? In so far as the plausibility of a Darwinian argu- 
ment turns upon the demonstration of conspicuous selective 
advantages. Class B adaptations are as amenable to Darwinian 
explanation as any other. It is indeed an explanation with 
many obscurities and shortcomings; but all I seek to emphasize 
is that none of them is peculiar to adaptations of Class B, i.e. 
peculiar to adaptations of the only kind for which a Lamarckian 
explanation is theoretically admissible. The adaptive value or 
•"selective advantage' of having developmentally prefabricated 
flexure lines is far from obvious; but so also is, for example, the 
adaptive value of many of the antigenic variants that deter- 
mine blood-group polymorphism and the incompatibilities 
revealed by grafting — differences upon which a Lamarckian 



interpretation has no bearing whatsoever. Flexure Hnes are 
mysterious, but not mysterious in any way that is particularly 
discreditable to Darwinism. 

This section may well conclude with a description of an 
important experiment in which Waddington (1952) has demon- 
strated the genetical pre-emption of a change originally brought 
about by environmental means. If fruit flies are subjected to a 
mild temperature shock shortly after pupation, a certain pro- 
portion develop without the cross veins that bridge the prin- 
cipal veins of the wings. Flies of this susceptible fraction were 
bred from, and their ofl'spring again shocked; the susceptible 
fraction again bred from, and so on. The proportion of suscept- 
ible flies steadily increased, as was to be expected; but from 
the twelfth generation onwards, the cross-veinless condition 
began to appear in flies which had received no temperature 
shock at all. Selection has thus, in eff'ect, converted an acquired 
into an inherited character diff'erence.* 

The gist of the foregoing argument is as follows. Darwinism 
and Lamarckism may be thought of as competing interpreta- 
tions of the origin of inherited character difi'erences in metazoan 
individuals. An examination of these character diff'erences 
shows that only a certain category, described as Class B, is 
open to a Lamarckian interpretation at all. But there is, on 
the one hand, no evidence to suggest that the Lamarckian 
interpretation is the correct one; and, on the other hand, 
Darwinism is no less competent to explain the origin of Class B 
adaptations than the origin of any other. 


The weak form of Lamarckism, which we have seen to be 
unobjectionable, is purely descriptive in intent; it merely 

■^ [Some of Waddington's more recent experiments are reported in 
Evolution, 10, p. 1, 1956.] 



describes a biological history of the origin of certain inherited 
character differences. The '"strong"' form of Lamarckism is the 
weak form strengthened (in the sense of being made more 
particular) by the categorical statement that the origin of 
acquired character differences is accompanied by the origin of 
adaptive genetical differences in the individuals in which they 
are induced. By an ''adaptive'' genetical change is only meant 
such a change as will reproduce the character difference 
originally elicited by the environment: the enlargement of a 
particular muscle by habitual use must be accompanied by 
such a genetical change as will entail the enlargement of that 
muscle. The qualification ''adaptive'' is therefore of central 
importance. That differences of environment or of ""treatment"* 
may bring about genetical transformations has not been in 
dispute since Muller"'s demonstration, now a quarter of a 
centur}^ old, of the mutagenic action of X-rays, and the number 
of physical and chemical treatments known to increase muta- 
tion rate is being steadily added to. 

Lamarckists do not suppose that adaptive genetical changes 
are completed within a single generation; the ''strong"' form of 
Lamarckism may therefore be expressed in such a way as to 
take this qualification into account: 

The repeated induction of character-differences within the 
lifetimes of individuals of successive generations is accompanied 
by a genetic change in each individual, the change being such as 
eventually to reproduce the character-difference elicited by en- 
vironmental stimuli even when those stimuli are withheld. 

It will be clear that the only acceptable evidence for Lam- 
arckian inheritance in the strong sense will be that in which 
the possibility of selection is scrupulously eliminated. This 
section will begin by a consideration of four examples of 
supposedly Lamarckian inheritance in higher animals, choosing 
the experiments on the grounds that they have been conducted 
with care and reported in sufficient detail to make an appraisal 



possible, and avoiding those in which there is a suspicion of 
corrupt advocacy. 

3a. lamarckian inheritance in 
higher organisms 

(i) The inheritance of eye-defects induced by specific atitisera. 
Guyer and Smith (1918, 1920, 1924), although not themselves 
'particularly interested in establishing or disestablishing any 
ism"*, claimed to have shown that eye defects induced in rabbit 
foetuses by the injection of pregnant does with anti-lens serum 
were reproduced in successive generations born of the affected 
rabbits. In a representative experiment, rabbits'* lenses were 
pulped and injected into chickens to elicit the formation of 
anti-lens precipitating antibodies. The antiserum so formed 
was injected into pregnant does. A small proportion of the 
offspring were born with eye abnormalities ranging from 
opacity and mis- shapenness of the lens to an apparently com- 
plete ''liquefaction''. These induced differences of eye structure 
were inherited, in roughly the manner of a Mendelian re- 
cessive, through both male and female lines. 

With variations that may have been significant, these claims 
were tested by three independent groups of workers (Finlay, 
1924; Huxley and Carr-Saunders, 1924; Ibsen and Bushnell, 
1931, 1934) with negative results. The findings of Guyer and 
Smith therefore remained in the penumbra of unexplained 
anomalies until Sturtevant (1944) proposed a prima facie 
genetical case for their acceptance. Following a train of 
thought started by M. R. Irwin and J. B. S. Haldane he argued 
that, in as much as there is in general a one-to-one correspond- 
ence between particular antigens and particular genes, an 
antigen may be *"a rather direct gene product** and may be 
imprinted with some of the structural specificity of the gene. 
'If a particular gene is responsible for the formation of a given 



antigen, there is a possibility that antibodies induced by this 
antigen may react with the gene/ In other words, the anti-lens 
serum, in addition to acting directly upon the foetal lens, may 
have altered in a genetically reproducible way the structural 
specificity of one or more ''lens genes\ Sturtevant refers to 
unpublished (and apparently still unpublished) evidence of 
R. R. Hyde in support of the original authors' claims. 

Guyer and Smithes experiments are plausible in a purely 
immunological sense, quite apart from the fact that they were 
done in a period when the authors could hardly have hoped for 
a genetical benediction. An '"anti-kidney'* or 'anti-mesenchyme'' 
immune serum would be expected to be quite ineffective, be- 
cause the immune bodies would be promptly absorbed by the 
corresponding maternal tissues and so denied access to the 
foetus. But the lens of adult rabbits is avascular; anti-lens 
antibodies should not therefore be absorbed by the mother but 
should be left free to act upon the vascularized lens of the 
foetus. Nor is there anv doubt that antibodies can reach the 
rabbit foetus — not through the placenta, as was formerly 
believed, but through the yolk sac (see Brambell, Hemmings 
and Henderson, 1951) Unfortunately, there is discrimination 
against antibodies ("heterologous antibodies'*) formed in an 
organism of a foreign species, and this, combined with the very 
decided toxicity of foreign serum as such, makes one regret 
that Guyer and Smith did not persevere with the experiments 
in which they tried to elicit anti-lens antibodies from the rabbit 

We must not, however, be led astray by speculations on 
whether or not the phenomena described by Guyer and Smith 

* [It has occurred to me, as a possible explanation of Guyer and Smith's 
positive results, that the lens preparations which they used as antigens 
may have been contaminated with bacteria. Bacterial antigens are now 
known to exert a powerfully 'adjuvant' action upon the production of 
antibodies by simple antigens, and this applies to auto-antibodies as well. 
See J. Freund, Advances in Tuberculosis Research, 7, p. 130, 1956.1 



could happen; the problem is whether or not they do happen, 
and the answer to this problem is at present open. 

(ii) The inheritance of learned behaviour differences in rats. 
McDougall argued that a fair test of Lamarckian inheritance 
should be one in which the acquired character difference repre- 
sented the outcome of an active and (in the everyday sense) 
'purposive"* response by the subject, and should be such that 
the results were open to quantitative assessment. He therefore 
studied the inheritance of the acquired ability of rats to 
learn one of alternative methods of getting out of a water trap. 

The trap was a water bath with a central entrance ramp and 
two exit ramps, one brightly illuminated and so wired as to 
give a tetanizing shock, the other dim but not electrified. The 
two exits were alternated to prevent the complication of the 
experiments by the learning of left-handed or right-handed 
habits of emergence. The rats came from the inbred stock of 
the Wistar Institute, and were divided into three groups of 
which two were bred from at random, or at least without avoid- 
able selection. The three groups were {a) untrained controls; 
(6) experimental rats that had been trained in the tank; and 
(c) rats which had been through the tank tests but which, 
instead of being bred from at random, were deliberately 
selected for breeding from those which showed the worst per- 
formances. The criterion of learning status was the number of 
tests that had to be given before an individual scored twelve 
correct choices of exit successively. 

The results of McDougalPs experiments, reported over a 
period of years in the British Journal of Psychology (1927, 1930, 
1938; McDougall and Rhine, 1934), were as follows. The tank- 
trained rats of group (h) improved in performance from a score of 
120 errors in the first generation to only 36 in the thirty- fourth 
generation of non-selective inbreeding. Unfortunately, the rats 
'negatively'' selected from the dullards of each generation 
(group c) improved from performance scores of 215 to 43 over 



the same period, and the untrained controls improved from 
149 to 102 over a period of only four years. 

Careful independent repetitions of McDougalPs work by 
Crew (1936) and Agar, Drummond and Tiegs (1935, 1948) 
failed altogether to confirm his empirical findings; they were 
not scrupulously exact repetitions, it is true, but embodied 
refinements that increased the precision of the experiments 
without in any way aff'ecting the principle of their design.* 
McDougalPs results are therefore on a somewhat different 
footing from those of Guyer and Smith. ^ What part could 
selection have played? In theory no part, for the experimental 
subjects had been inbred for a sufl^icient number of generations 
to justify the prevailing theoretical assumption that they were 
genetically uniform and homozygous. In practice, this pre- 
sumption seems to have been unduly optimistic: Loeb"'s work 
(1945) on the transplantation of tissues between members of 
the highly inbred Wistar strain of rats revealed incompati- 
bilities that can only have been due to flagrant heterozygosity. 
Guinea-pigs and mice, by contrast, become completely tolerant 
of grafts transplanted between members of an inbred line after 
a much less prolonged regimen of inbreeding. McDougalPs 
stock may, then, have been more heterogeneous than is usually 
supposed — and, as Drew (1939) has made clear in his admirably 
succinct review, there is plenty of evidence that differences of 
intelligence in rats, as measured by maze performances, are 
perfectly amenable to selection. 

It may of course be argued that McDougalPs adverse selec- 

1 Haldane (1951) makes the comment that McDougall's colleague and 
pupil, Rhine, was conducting experiments in paranormal cognition in the 
same laboratory, and points out the inconsistency of presenting evidence 
in favour of paranormal cognition in human beings without taking into 
account its effect on the outcome of such experiments as McDougall's. 

* [The final report on this long and important experiment has now been 
published: W. E. Agar, F. H. Drummond, O. W. Tiegs and M. M. Gunson, 
Journal of Experimental Biology, 31, p. 307, 1954.] 



tion experiments prove that his results could not have been due 
to inadvertent selection. Unfortunately, the results from group 
(c) raise the new difficulty that improvement was more striking 
in the line perpetuated by dullards than in the unselected 
experimental stock; and there appears to have been a general 
secular improvement in the group (group a) which had not 
been exposed to the tank tests at all. McDougalPs case must 
stand or fall by the empirical results, and these have not been 

(iii) Melanism in moths. ""The spread in industrial districts of 
melanic forms of Lepidoptera is . . . one of the most consider- 
able evolutionary changes that has ever actually been witnessed' 
(Ford 1940). The change is widespread and has been rapid. 

It was argued by Heslop Harrison (1926, 1928) that melanism 
is an induced and inheritable adaptive change: food plants in 
industrial areas were held to be contaminated by metallic 
fumes, and Harrison claimed to have induced the formation of 
melanic mutants by feeding larvae of the moth Selenia bilunaria 
on hawthorn leaves which had absorbed small quantities of 
salts of manganese and lead. Hughes (1933; cf. also Thomsen 
and Lemche, 1933) repeated Harrison''s experiments with six 
generations comprising 3265 individual moths and found no 
melanic forms among the treated or the untreated; he adds that 
manganese salts are present in normal plants and are not 
present to excess in plants of industrial areas. 

There is a clear-cut alternative explanation of the spread of 
melanism in moths, for which we are indebted to Ford. Melan- 
ism is a mutant of regular occurrence in many species of 
Lepidoptera from non-industrial areas; there is therefore a 
clear case for supposing that such mutants have been selected 
for their superior viability in the smoke-stained countryside of 
industrial districts. Indeed, the experience of many workers 
has been that certain melanic mutants are tougher and more 
viable than the ordinary paler forms; presumably they have 



failed to spread to non-industrial areas because their advantage 
in toughness is more than outweighed by their greater con- 

(iv) Sladdeii's experiments on the inherita?ice of altered food 
habits in stick-insects. These are perhaps the best of the experi- 
ments that purport to demonstrate Lamarckian inheritance; 
all sorts of genetical complications are avoided by the fact that 
reproduction in the subject species is parthenogenetic. Sladden 
(1934, 1935; Sladden and Hewer, 1938) studied the inheritance 
of the acquired ability of stick-insects of the species Dixippus 
morosus to subsist upon ivy instead of their normal diet, privet. 
The life cycle in this species is 9-10 months long, and somewhat 
more than 500 eggs are produced by each individual. 

The insects feed at night, and must feed every night. The 
alternative foods were offered for consumption in such a way 
as to provide a reliable measure of their degree of accept- 
ability. In the ''presentation test\ ivy and privet were offered 
on alternate nights, the privet being necessary to keep the 
insects alive if they failed to eat sufficient ivy. Acceptability 
was measured by the number of trials necessary before the 
final acceptance of ivy. In the 'preference tesf ivy and privet 
were thrice offered simultaneously: the result was scored as 
'ivy preference** if ivy was chosen on all three occasions, and so 
for privet; otherwise the result was held to be unindicative. 

After six generations there was a clear-cut increase in the 
} acceptability of ivy, but it is noteworthy that a high proportion 
of this increase occurred in the first generation after the first 
presentation of ivy. There is also an echo of the difficulties that 
bedevil the interpretation of McDougalPs work, in that the 
control insects, reared upon privet throughout, also showed a 
distinct increase in preference for ivy. Sladden''s own inter- 
pretation of this finding, which turns upon seasonal changes of 
food preference, is unconvincing. 

These are good experiments: the facts are well set out and 

G 97 


their truth is not in question. Thorpe (1938, 1939) has however 
suggested an alternative and rather unexpected interpretation 
based upon the fact that insects are susceptible of a high degree 
of olfactory conditioning, in the sense that odours normally 
distasteful to adults may be acceptable if larvae are exposed to 
them early enough. For example: the ichneumon fly Nemeritis 
canescens normally lays its eggs in the Mediterranean flour 
moth Ephestia kuhniella^ and is strongly attracted by the smell 
of its normal host. It does not normally lay eggs in, and is not 
normally attracted by, the smell of the related wax moth 
Meliphora. But if the ichneumon flies have been deliberately 
reared in Meliphora, or have been exposed to its larvae shortly 
after emergence from the cocoon, then they do show a strong 
attraction to Meliphora. This transformation of host preference 
was complete in one generation; ten successive generations of 
rearing on Meliphora did not increase it. 

With this and other evidence of similar import in mind, 
Thorpe therefore suggests that, in Sladden''s experiments, some 
olfactory emanation from ivy caused a conditioning which 
increased its acceptability to stick-insects. Enough might arise 
from the egg to condition the newly hatched nymphs, particu- 
larly if their first food is egg-shell. This does not account for 
the progressive increase in the acceptability of ivy over six 
generations, but as the greater part of this increase occurred 
after the first generation, and as there was some increase in 
tolerance by the controls, it is difficult to regard this as a grave 
shortcoming of Thorpe''s explanation. 

These four examples inspire one with no confidence in the 
applicability of the Lamarckian scheme of inheritance to higher 
animals. Two are susceptible of clear-cut alternative explana- 
tions; a third, McDougalPs, is open to question on the grounds 
of empirical fact; and the fourth, that of the inheritance of 
induced eye defects, is urgently in need of reinvestigation. I am 
not aware of any experiments that have a greater claim upon 



our attention than these four, though of many which have less. 
It is therefore the generally held view that the case for Lam- 
arckian inheritance in metazoa is unproven. 

3b. lamarckian inheritance in 

There can be no doubt that a mode of inheritance which 
satisfies the definition with Avhich this section began is demon- 
strated by non-cellular organisms. In such organisms the entire 
body substance participates in the act of reproduction, so that 
the argument against Lamarckism which turns on the physical 
inaccessibility of the germ plasm to environmental influences 
loses much of its force. 

Two examples will be cited. It will be as well to say at the 
outset that they are founded upon experiments of exemplary 
design and scrupulous care of interpretation, and are thus 
wholly free from the taint of muck-and-mystery speculation 
for which so many Lamarckists have an unfortunate predilec- 
tion. The intelligibility of the experiments, the fullness and 
clarity of their exposition, and the hope they offer of rigorous 
scientific interpretation must not, however, allow us to infer 
that the modes of inheritance they reveal cannot really be 
Lamarckian. There is nevertheless an excellent informal reason 
why to describe them as Lamarckian is singularly pointless. 
When a phenomenon apparently sui generis is shown to belong 
to some wider class of phenomena — as, for example, when 
allergies are shown to belong to the general class of immunity 
reactions or, to go far back in zoological history, when par- 
thenogenesis is shown to be a variant of sexual reproduction — 
then much is gained; for the phenomenon so classified is at once 
given access to and support from a large and solid body of 
reasoning and experiment which can be used to suggest new 
pathways of research and new schemes of interpretation. With 



the examples to be considered here, it is far otherwise: to 
describe them as Lamarckian is to open the door, not to a 
bright theoretical illumination but to a fog of undisciplined 
fancies. If I persist in calling them Lamarckian, it is because 
Lamarckism happens to be the subject of the present essay. 

The two examples have been chosen for their familiarity to 
the present writer; but the whole subject of cytoplasmic in- 
heritance in micro-organisms and its bearing upon the problems 
of cellular differentiation in metazoan development have been 
comprehensively reviewed by Professor B. Ephrussi in his 
recent Withering Lectures (1952), which should be referred to 
for fuller information.* 

(i) The inheritance of acquired resistance to antisera in Para- 
mecium aurelia. Paramecia may be immobilized or killed by the 
incorporation into their culture-media of an antiserum formed 
by injecting suspensions of whole individuals into rabbits. If 
Paramecia are cultivated in sub-lethal concentrations of anti- 
serum, their progeny acquire a resistance to its action under 
conditions which (it is now known) completely exclude the 
mere selection of the more resistant forms for propagation 
Resistance so acquired is retained for many generations of 
asexual fission — in some varieties, through sexual fission as 
well — in the complete absence of the stimulus which originally 
brought about the transformation. Evidently the antiserum 
has initiated a heritable change. 

These phenomena have been studied in recent years by 
Bernheimer and Harrison (1940, 1941), Harrison and Fowler 
(1945, 1946) and Kimball (1947); most of our information, 
however, derives from the detailed and systematic genetical 
analyses of Sonneborn (reviews 1949, 1950) and more recently 
of Beale(1952).t 

"^ [Nucleo-cytoplas7nic Relations in Micro-organisms, Oxford, 1953.] 
t [The most comprehensive modern summary of this work is The Genetics 
of Paramecium aurelia, by G. H. Beale, Cambridge, 1954.] 



In very brief outline the evidence may be summarized thus. 
An individual Paramecium aurelia belongs to a variety — 
essentially a species; to a 'type\ which is an assembly defined 
by mating compatibilities and so equivalent to a sex; and to a 
stock. A stock is the progeny of a single homozygous individual. 
Within a stock, an individual may display one and (except 
while a tranformation is actually afoot) only one of a distinct 
set of surface antigens defined and labelled by their power to 
elicit specific antibodies from the rabbit. Diff'erences of anti- 
genic composition between the individuals of a stock are herit- 
able, but they depend upon diiferences of cytoplasm and not 
upon differences of nuclear genes. Different stocks are dis- 
tinguished by different combinations of the antigenic char- 
acters that may be displayed by their constituent members, 
and these differences of antigenic potential are governed by 
differences of nuclear genes. (It seems likely that the same gene 
loci are represented in all the stocks of a given variety, and 
that differences of antigenic composition between stocks 
depend upon different representations of the alleles of these 
loci.) Within a given stock, however, it is an inherited cyto- 
plasmic difference that discriminates between the range of 
antigenic possibilities governed by the prevailing nuclear 

If an individual or an assembly of similar individuals is 
exposed to a sub-lethal concentration of the antibody directed 
against the prevailing surface antigen, a heritable transforma- 
tion is brought about, in consequence of which the prevailing 
antigen is replaced by another member of the set characteristic 
of the stock. The effect of the transformation is to confer 
resistance to an antibody upon the progeny of an individual 
which was formerly susceptible to it. It is of some importance 
that such transformations may also be brought about, though 
(so far as present knowledge goes) more slowly, by a variety of 
'non-specific'' stimuli such as changes of temperature or 



nutritional status, or by treatment with enzymes (Kimball, 

It is clear that the cytoplasm of Paramecia is malleable in a 
way completely foreign to our conception of the propagation 
system of the chromosomes, and that this malleability endows 
them with what is, in effect, a cytoplasmic genetic memory. 
We shall not delay with interpretations of the mechanism of 
the adaptive response, except to say that all turn upon the idea 
of an intracellular competition, whether between self-perpetu- 
ating cytoplasmic particles or between reaction sequences that 
are mutually inhibitory and so mutually exclusive. Such an 
interpretation gives point to Hinshelwood's comment that 
inheritance which is Lamarckian in terms of cells should be 
described as Darwinian at the level of cellular ingredients (see 
Section 1). 

(ii) Adaptive transformations in micro-organisms. The ""train- 
ing** of micro-organisms is an old story in bacteriology, but it is 
only in quite recent years that it has been seen to have an 
educational import for zoologists as well as for bacteria. 

Bacteria may be trained to use lactose or glycerol instead of 
glucose as a source of carbon; nitrates instead of atmospheric 
oxygen; ammonium salts instead of amino-acids as a source of 
nitrogen; and so on. They may also be trained to resist anti- 
biotics and other growth inhibitory agents to which they were 
at first susceptible. 

The interpretation of the mechanism of these changes is 
complicated by two facts: (a) bacteria are too small for it to be 
possible to study their individual histories in sufficient detail, 
so that the behaviour of individuals must be inferred from the 
behaviour of bacterial populations; (6) the gene is not known 
as a unit of segregation but only as a unit of mutation. The 
modern analysis of recombination phenomena in viruses and 
bacteria (Delbriick and Bailey, 1946; Tatum and Lederberg, 
1947) will no doubt correct this second shortcoming in due 



course. But as matters stand at present, the interpretation of 
Hraining"* adaptations is highly controversial, the controversy 
being between those who maintain that training is secured both 
by population selection and by heritable transformations of 
individual cells, and those who maintain that only the former 
mechanism is at work. 

It may be said at once that numerous studies of the highest 
exactitude have made it clear that the differential survival of 
genetic variants is a ubiquitous property of populations of 
micro-organisms, and the truth of this proposition is not 
therefore in dispute. At the same time, the prolonged and 
exact experiments of Hinshelwood (1946; cf. also Baskett and 
Hinshelwood, 1951), expressly designed to discriminate be- 
tween selection and adaptation, show that the adaptive trans- 
formation of individual cells could well be a capital factor 
in the training responses of bacterial populations.* The two 
mechanisms are no more incompatible in bacteria than in 
Paramecia; indeed, there is a formal analogy between antigenic 
transformations in the latter and training responses in the 
former. For example, Kilkenny and Hinshelwood (1951) have 
compared the adaptation of three strains of the yeast Saccharo- 
myces cerevisiae to the use of galactose: one adapted itself 
promptly, a second slowly and the third not at all. These diifer- 
ences of adaptive potential were inherited according to the 
ordinary rules of Mendelian segregation, but within each strain 
adaptation, if it occurred, was brought about by transforma- 
tions of individual cells. 

Hinshelwood, it may be noted, has more than once insisted 
that any differences of adaptive capacity between individuals 
are bound to be inflated by selective forces, and the com- 
patibility of 'adaptive** and ""selective' explanations may be 
illustrated by reference to Paramecia. If Paramecia of mixed 

* [Many later experiments by Hinshelwood 's school have been published 
in Series A of the Proceedings of the Royal Society.^ 



antigenic types were to be subjected under carefully calculated 
conditions to the action of an antiserum directed against only 
one, then the resulting transformation of the assembly con- 
sidered as a whole would represent the outcome of two pro- 
cesses: (a) the continued propagation, at first numerically 
favoured, of individuals lacking the antigen against which the 
antiserum was effective, and (b) the transformation of formerly 
susceptible individuals into different antigenic types. Denied 
the use of a microscope, it would have been very much more 
difficult — it has not in any case been easy — to distinguish 
between the contributions of the two processes, or, indeed, to 
be certain that both occurred. It may be agreed, then, that 
micro-organisms show heritable adaptive transformations of 
individual cells, and that these are superimposed upon, and act 
within limits governed by, the Mendelian mechanism of 
genetic inheritance. 

There seems to be no great mystery about the significance 
for Paramecium itself of the type of inheritance illustrated so 
well by antigenic transformations. Paramecium is a very 
vulnerable organism, with a very short interval between suc- 
cessive generations. It is to its advantage not only to be able 
to transform itself in the face of adverse circumstances but 
also to be able to take several generations to do so, if the 
change cannot be accomplished in one. Moreover, the com- 
pleted change is passed on ready-made to succeeding genera- 
tions; it would be a hopeless arrangement, from Paramecium's 
point of view, to start from scratch in each generation. 

But what is the wider significance of the fact that Para- 
mecium enjoys two systems of inheritance: the nuclear or 
gene-determined, and the cytoplasmic or gene-limited? It is 
possible that of the two collateral systems of inheritance dis- 
played by protozoa the nuclear system persists in the mech- 



anism which determines the differences between the zygotes of 
higher organisms, and the cytoplasmic system, in a highly 
regimented form, persists in the mechanism which gives rise to 
differences between the cells that descend by mitotic division 
from the zygote. In other words, cytoplasmic and nuclear 
systems of inheritance live side by side in micro-organisms 
because heredity and development have not yet sorted them- 
selves apart; a lineage of protozoans has something in common 
both with a lineage of higher organisms and with the lineage 
of cells which arises from the zygote of each one. It may there- 
fore be, as Sonneborn has long insisted, that in studying the 
cytoplasmic hereditary mechanisms of protozoa one is attack- 
ing not indeed the problem of embryonic differentiation itself 
but the first and perhaps most vulnerable outpost of its 
remarkably stubborn defences. 


Abercrombie, M. (1952). New Biology, 13, p. 117. 
Agar, W. E., Drummond, F. H., and Tiegs, O. W. (1935). /. exp. 
Biol, 12, p. 191. 
(1948). Ihid., 25, p. 103. 
Baskett, A. C, and Hinshelwood, C. (1951). Proc. Roy. Soc. B, 139, 

p. 58. 
Beale, G. H. (1952). Genetics, 37, p. 62. 

Beer, G. R. de (1951). Embryos and ancestors. 2nd Ed., Oxford. 
Bernheimer, A. W., and Harrison, J. A. (1940). /. Immunol., 39, p. 73. 

(1941). Ibid., 41, p. 201. 
Billingham, R. E., and Medawar, P. B. (1948a). Heredity, 2, p. 29. 
(19486). Brit. J. Cancer, 2, p. 126. 
(1950). /. Anat., Lond., 84, p. 50. 
Brambell, F. W. R., Hemmings, W. A., and Henderson, M. (1951). 

Antibodies and embryos. London. 
Clark, W. E. Le Gros (1952). The tissues of the body. 3rd Ed., 

Crew, F. A. E. (1936). J. Genet., 33, p. 61. 



Delbriick, M., and Bailey, W. T. (1946). Cold Spring Harb. Symp. 

Quant. Biol., 11, p. 33. 
Drew, J. S. (1939). Nature, 143, p. 188. 
Finlay, G. F. (1924). J. exp. Biol, 1, p. 201. 
Fisher, R. A. (1930). The genetical theory of natural selection. 

Ford, E. B. (1940). Ann. Eugen., 10, p. 241. 
Guyer, M. F., and Smith, F. A. (1918). J. exp. Zool., 26, p. 65. 

(1920). Ibid., 31, p. 171. 

(1924). Ibid., 38, p. 349. 
Haldane, J. B. S. (1951). XIX Congr. Int. Philosophie des Sciences, VI, 

Biologic, p. 39. 
Harrison, J. A., and Fowler, E. H. (1945). /. Immunol., 50, p. 115. 

(1946). J. exp. Zool, 101, p. 425. 
Harrison, J. W. H., and Garnett, F. C. (1926). Proc. Roy. Soc. B, 

99, p. 241. 
Harrison, J. W. H. (1928). Proc. Roy. Soc. B, 102, p. 338. 
Hinshelwood, C. N. (1946). The chemical kinetics of the bacterial 

cell. Oxford. 
Hughes, A. W. McK. (1932). Proc. Roy. Soc. B, 110, p. 378. 
Huxley, J. S., and Carr-Saunders, A. M. (1924). /. exp. Biol., 1, p. 215 
Ibsen, H. L., and Bushnell, L. D. (1931). /. exp. Zool., 58, p. 401. 

(1934). Genetics, 19, p. 293. 
Jones, F. Wood (1943). Habit and heritage. London. 
Kimball, R. F. (1947). Genetics, 32, p. 486. 
Kilkenny, B. C, and Hinshelwood, C. (1951). Proc. Roy. Soc. B, 

139, p. 73. 
Loeb, L. (1945). The biological basis of individuality. Springfield, 

McDougall, W. (1927). Brit. J. Psychol., 17, p. 267. 

(1930). Ibid., 20, p. 201. 

(1938). Ibid., 27, p- 321. 
McDougall, W., and Rhine, J. B. (1934). Brit. J. Psychol., 24, p. 213. 
Medawar, P. B. (1951). Ne?v Biology, 11, p. 10. 

(1952). Ibid., 13, p. 116. 
Sladden, D. (1934). Proc. Roy. Soc, B 114, p. 441. 

(1935). Ibid., 119, p. 31. 
Sladden, D., and Hewer, H. R. (1938). Proc. Roy. Soc. B, 126, p. 30. 
Sonneborn, T. M. (1949). Ann. Rev. Microbiol., 3, p. 55. 

(1950). Heredity, 4, p. 11. 



Sturtevant, A. H. (1944-). Proc. Nat. Acad. Sci., 30, p. 176. 
Tatum, E. L., and Lederberg, J. (1947). /. Bad., 53, p. 673. 
Thomsen, M., and Lemche, H. (1933). Biol. Zbl., 53, p. 541. 
Thorpe, W. H. (1938). Proc. Roy. Soc. B, 126, p. 370. 

(1939). /6/d., 127, p. 424. 
Waddington, C. H. (1952). Nature, 169, p. 278. 
Woodger, J. H. (1952). Biology and language. Cambridge 


The Pattern of Organic 
Groivth and Transformation 

'Growth'' is a word of notorious imprecision, but it stoutly 
defies semantical reform. It may mean increase of length, area, 
weight or volume; it may mean the act or accomplished fact of 
reproduction, i.e. increase of number; or it may simply mean 
development — the adverb is not well chosen — with all that 
development implies of increasing complexity and elaboration. 
I shall restrict growth here to its simplest meaning, change of 
size, but I shall consider also the changes of shape which are 
the outcome of inequalities in the rate of change of size. 

Organic growth is not a process of accretion, nor does it 
build upon an enduring frame. The molecular fabric of the 
body enjoys no substantive permanence whatsoever, a truth 
which came to be known in the following way. 

The body makes no distinction between the common ele- 
ments and their various mutants; the natural isotopes of 
nitrogen and carbon, which have atomic weights of 15 and 13 
instead of 14 and 12, or the radioactive isotopes of sodium (24) 
or carbon (14) which arise by gaining neutrons or losing 
protons, are exchanged indifferently for their common or 
parental forms. The administration of compounds containing 
isotopes distinguished by their mass or radioactivity has there- 
fore made it possible to trace atoms in their passage through 
the body, and so to reveal the constant exchange of its mole- 
cular ingredients for new arrivals from the world outside. Even 



teeth and bone are the subjects of a restless atomic tran- 
substantiation. It is only the form of the body, the system of 
preferred stations for the inward-bound replacements, that 
achieves any kind of permanence at all. 

Superimposed on these exchanges are the processes which 
make good the constant wastage of effete or expended cells. 

Fig. 6. Male beetles of the species Euchirus longimanus, 
illustrating how the proportions of an organism may 
change with its absolute size. The length of the fore-limbs 
is grossly out of proportion to the length of the body. 

Pounds of dead cells in the form of scurf and its several 
variants (hair, horn, nails, claws) are parted with in a lifetime. 
The living outer layer of human skin renews itself completely 
about once a month, or about 100 times in a proverbial seven 
years. Red blood corpuscles live only about 120 days; at least 
some lymphocytes appear to be excreted through the walls 
of the intestine; a small proportion of the finest nerve fibres 
and blood vessels is probably always in course of disintegration 
and therefore always in course of being formed anewj Replace- 



ments of this kind are part of the ordinary maintenance charges 
of the body: they are not accompanied by any net change of 
size. But some forms of wastage are integral with the act of 
growing. Bony tubes and boxes Kke the long bones of the legs 
and the cranium are hollowed out on the inside in the course of 
growing larger. The only growth which is purely additive or 
accretionary is that of which the product takes no further part 
in the physiological activity of the body, as with shells or hair. 
The idea that the growth of organisms can be likened to, for 
example, the growth of houses is not acceptable even in the 
roughest first approximation. The two processes have nothing 
in common at all. 

In spite of the complexity of growth, its outcome, as we 
measure it, may be comparatively simple, and in later para- 
graphs I shall set out some of the quantitative rules to which 
growing animals conform. The measurements I shall refer to 
tell one no more and no less about growth than could be 
learned of the mechanism of respiration by measuring the 
composition of inspired and expired air, or of a firm's 
method of conducting business by contemplating a single 
figure representing its annual net loss or gain. In all such 
cases we have to do with measuring the final outcome of 
covert processes of formidable complexity. The measurements 
are not very deeply informative, but the information which 
they contain is indispensable. 


The largest adult mammals are about 50 million times larger 
than the smallest. A fully grown blue whale weighs about 
2 X 10^ grammes; one of the smallest mammals, the long-tailed 
shrew Cryptotis parva parva, weighs only about four. Even 
when studied under conditions ''particularly conducive towards 
repose** this shrew ate its own weight of worms and insects 



daily, and would have died of starvation if food had been 
withheld for as little as twelve hours. 

The scale of sizes to be found in mammals is not exceptional. 

Fig. 7. Differential growth of the skull of the baboon 
as seen from the under surface and inside view: 1, 2, and 
3 are from newborn, juvenile, and adult females, and 4 
from an adult male. 

The Gobiform fish Misticthys is about half an inch long when 
fully grown and could hardly weigh 1 /250th of an ounce. The 
basking shark (not the largest fish) is known to reach 29 feet 



in length and to weigh four tons. Dr Harrison Matthews has 
given excellent reasons for supposing it to be viviparous, 
though no pregnant specimen has yet been found. The Japan- 
ese spider crab may have a claw span of ten feet in extension, 
but the smallest crustaceans are little more than an animated 
sea dust in the surface waters of the ocean. The smallest beetles 
and fairy flies are about 1 /100th of an inch in length. The 
largest squids are 90 feet long and have eyes as big as saucers. 
It is not possible to say exactly why animals of a particular 
species should have come to be of a particular size. The sizes 
and growth rates of animals are functionally in gear with all 
the other parameters that define their way of living — their rate 
and manner of reproduction, their behaviour, habitat, enemies 
and food. But it is sometimes possible to see why animals 
cannot be very much larger or smaller than they are. One very 
general restraint turns on a metrical truism recognized by 
Spencer — namely, that in a body which is symmetrically en- 
larging, the volume increases as the cube of the linear dimen- 
sions, and the surface area as the square. To multiply length 
tenfold is to increase surface area a hundredfold and volume a 
thousand times. In small mammals the ratio of surface area to 
volume, and therefore the relative rate of loss of heat, is much 
greater than in large mammals. The smallest mammals eat 
almost continuously to make good the loss of heat and could 
not very well be smaller. At the other extreme, the elephant is 
approaching the upper limit of size for an agile and wholly 
terrestrial animal. Limbs are roughly speaking strong in pro- 
portion to their cross-sectional areas but support a weight that 
is proportional to volume. The legs of elephants must of 
necessity be stouter and more pillar-like than the legs of 
horses; indeed, elephants can be extrapolated for fancy to a 
size at which one would be lucky to see daylight between their 
legs. My colleague Mr Majrnard Smith estimates that the 
upper limit of the weight of a flying vertebrate must be about 



30-40 lb., because the power needed for flight increases more 
rapidly than as the cube of the linear dimensions, and therefore 
out of proportion to the muscles which provide the motive 
power. Angels, paradoxically, could therefore not be airborne, 
as Professor Haldane pointed out some quarter of a century ago. 

Spencer''s Law is more revealing in its actual breach than in 
its theoretical observance. Surface area keeps pace with the 
volume it ministers to by folding and subdivision. The walls of 
the intestine are deeply folded; the walls of the lungs are a 
multitude of fine sacs. The cross-sectional area of the blood 
vessels is thought to increase about 800-fold in the passage 
from the great vessels by the heart to the capillaries of the 
tissues. The five million red blood cells in a cubic millimetre of 
human blood offer a surface area 170 times greater than that 
of a single corpuscle of the same shape and total mass. Spencer''s 
Law is also flouted by physiological adaptation. If the problem 
of conserving heat is so acute for the smallest adult mammals, 
how do their newborn babies cope, which are so much smaller 
still, and hairless? The short answer is that they do not. New- 
born mice come to no harm by being left for an hour or two in 
a refrigerator. Their metabolism is such that they are highly 
resistant to the effects of being chilled. 

Another restraint is that which is set by the tempo of 
diffusion. In the simplest case a diffusing substance penetrates 
to a distance proportional to the square root of the time during 
which it has been diffusing. Distance can therefore only be 
bought at a disproportionate cost of time, a state of affairs 
which sets definite limits to the permissible shapes of cells. All 
active cells or one-celled animals which are large are tubular 
or flattened, except when like yolky eggs they derive energy 
from stores of food inside. 

I hope the foregoing discussion will have distracted attention 
from the fact that we are very ignorant of the actual and 
present influences which govern the size-distribution and 

H 113 


growth of any wild animal living under natural conditions. 
Fish are the only wild animals for which we are approaching a 
predictive ecological theory of growth rate and size frequency. 
Nor can this be counted a triumph of abstract scholarly enter- 
prise. The pressure of necessity is behind it; it is less because 
fish are edifying than because they are edible that we know as 
much as we do. 


In spite of the compass and complexity of growth, and the 
great variety of different processes that contribute to increase 
of substance, the passage from germ to adult is an orderly and 
predictable process. What rules of order does it conform to, 
and upon what reasoning is prediction based? 

It is one thing to devise empirical formulae which describe 
the growth of the members of one particular species; that is 
simply a matter of measuring the growth of a sufficient number 
under conditions sufficiently well defined. It is quite another 
matter to try to frame general laws of growth which the 
majority of animals are expected to conform to, and biologists 
have set about the problem in two entirely different ways. 

Some have attempted to arrive at Laws of Growth deduct- 
ively, starting with certain deceptively inoffensive axioms about 
the conduct of metabolism and ending with theorems that 
purport to describe the way in which all animals grow. I believe 
that this approach must be classified at present as a scholarly 
indoor pastime; that it may sometimes lead to acceptably 
accurate answers is only marginal evidence of the truth of the 


The other way to go about it* is to proceed inductively, by 

* [The inductive approach is considered in more detail in my article on 
'Size, Shape, and Age', in Essays on Growth and Form, ed. W. E. Le Gros 
Clark and P. B. Medawar (Oxford, 1945.)] 



recording as many instances of gro\\i:h as possible and trying 
to find out the properties they share in common. May we not 
aiRrm, for example, that animals increase in size as they grow 
older, until growth ceases altogether? We may, of course, but 
only if room is left for reservations. All animals grow smaller if 
undernourished — a trivial exception — but some animals (like 
flatworms, nemertines and colonial sea-squirts) ''de-grow'* with a 
deep-seated anatomical retrogression and may even revert to 
an embryonic level of simplicity. Negative growth of this kind 
has a special adaptive value. It is not a significant violation of 
the law of general increase because negative growth is not a 
reversal of the processes that led to enlargement, as if meta- 
bolism has simply been engaged in a reverse gear. Nor is it a 
significant exception that men and women are shorter in old 
age than in the physical prime. It is indeed so, and Dr Morant 
is satisfied that this shrinkage is not just an actuarial artifact 
due to an earlier death of taller people, nor to the fact that the 
older people we measure to-day were born longer ago than 
their juniors and therefore in perhaps less propitious times for 
growing. (Morant finds that Englishmen a hundred years ago 
reached the same maximum height as they do at present, but 
took about five years longer to achieve it.)* Loss of height is 
probably due to a shrinkage of intervertebral discs. This, too, 
is not a reversal of synthetic processes; and it may be observed 
that the luxury of living to an age at which one can indulge in 
physical deterioration is an artificial by-product of domestica- 
tion, and a state of afi'airs that has no parallel in the world of 
animals at large. 

It has long been recognized that biological growth is multi- 
plicative in style, and not accretionary or additive. That which 
results from biological growth is itself endowed with the power 
of further growing. In the general case the progeny of a cell 

* [Dr J. M. Tanner has since told me that there is clear evidence of a 
genuine secular increase in height as well as in growth rate.] 



(or chromosome or organism) which has divided into two are 
themselves capable of division, and so in turn their issue. 
Accretionary products like shells and hair are made by living 
cells which grow in the organic style: all additive growth is 
subsidized by acts of multiplication. 

A lineage of cells that perpetuated itself without loss by 
repeated binary divisions would of course increase in numbers 
in an exponential or geometrical progression. In real life, not 
even bacteria will increase at such a rate for long. Their growth 
is restrained by a variety of density-dependent factors, like the 
accumulation of inhibitory waste products or the exhaustion 
of the supply of food. Nevertheless, growth by continuous 
compound interest is the norm for all living systems. It is 
departure from exponential growth that calls for comment and 
explanation, just as with departure from uniform motion in a 
straight line. No moving object left to itself will persevere in 
constant linear motion, and no real organism will groAv at a 
constant specific rate. The former circumstance no more 
derogates from Newton^'s First Law of Motion than the latter 
from what is sometimes called the Law of Malthus. What we 
must ask is, in what way does the growth of organisms depart 
from that regimen of continuous compound interest by which 
they are theoretically empowered? 

No one has yet improved upon the answer given by the 
American anatomist Minot. Consider a sum of money invested 
at a rate of compound interest which, instead of remaining 
constant, falls; and let the interest be (say) 10 per cent in the 
first year, 9 per cent in the second, and 8-1 per cent, 7*3 per 
cent, 6'Q per cent ... in successive years thereafter. The sum 
does indeed grow at compound interest, but the rate of interest 
falls progressively at a rate which progressively falls. This is 
the organic style of growth as Minot saw it. A living system 
progressively loses its power to multiply its substance at the 
rate at which that substance itself was formed. Put otherwise, 



the specific acceleration of growth is always negative, but it 
climbs towards zero as growth proceeds. It is only superficially 
a paradox that deterioration is faster in young animals than 
in their elders. Almost all metabolic processes go faster in 

Fig. 8. Line drawings in side view of two related 
species of small marine fish allotted to different genera, 
Argyropehcus (left) and Sternoptyx (right). Plotted on a 
changed system of co-ordinates, the outline of the one 
gives an excellent approximation to the outline of the 

youth than in maturity, and the processes which slow down 
physiological activity are no exception. We are all moving 
towards our graves, but none so fast as they who have farthest 
still to go. 


Change of size is almost always accompanied by transforma- 
tion; growth by proportionate enlargement is very rare. We are 
not born as miniature adults. Inspected through a magnifying 
glass, a child does not look like a backward adult, but simply 
like an uncommonly large child. 

The shape that is characteristic of adults is governed by 
spatial inequahties of growth rate, i.e. growth that goes at 



different rates in different parts of the body. Dr J. S. Huxley 
has made a special study of these inequalities, and two ex- 
amples of change of shape cited in Problems of Relative Growth 
are shown in figs. 6 and 7. Adults of different but related 
species acquire their distinctive shapes because they conform to 
different but related rules of transformation. Some of their 
end results are shown in neighbouring figures (8, 9). These 
figures are taken from D''Arcy Thompson''s classical work On 
Growth and Form^ and more will be said of Thompsonian pro- 
jections later. 

The form of an object, unlike its size, cannot be expressed 
by a scalar quantity, a simple number. No child was ever 2*5 
Thompsons in form. Form must be expressed by a correlated 
system of vectorial measurements, i.e. measurements which 
take account of the disposition of the measured lengths in 
space. But although shape is in a purely metrical sense in- 
definable, change of shape is not. Consider a lantern slide 
thrown on a screen that lies in its normal position at right 
angles to the optical axis of the projector. When the screen 
tilts one way or another, the cone of light is cut at different 
angles and the image is accordingly transformed. The nature 
and degree of the distortion can be expressed with mathe- 
matical exactness. No matter how complex the pattern of the 
image, its change of shape can be accurately defined. 

Huxley''s method of assaying change of shape in development 
is to measure the growth rate of one part of the body in terms of 
the growth rate of another. If transformation is an orderly 
process, the two sets of measurements will vary in dependence 
on each other. In the simplest case, not uncommon but by no 
means universal, the parts in comparison multiply their sizes 
in a constant ratio: the size of one is a fixed multiple of the size 
of the other when the size of the other is raised to a constant 
power. Proportions alter, therefore, but alter in geometrical 
progression. It is only when the ratio or power is unity that the 



proportions of the growing parts stay constant, and this, as I 
have said, is rare. 

It follows that just as a growing animal must traverse all 

Fig. 9. The carapaces of crabs of six different but 
related genera, showing how particular differences of 
form may be expressed as the outcome of a general 
process of orderly spatial transformation. 



intermediate sizes before it attains to adult weight or stature, 
so it must traverse a spectrum of intermediate conformations 
before it can reach its adult shape. A particular shape can only 
be '■fixed*' if growth itself comes to a standstill or if the differ- 
ential growth-ratio settles down to unity, so that later growth 
entails symmetrical enlargement. The size and shape of an 
animal must therefore be nicely correlated. The advantages of 
being larger may be offset by unwieldy or otherwise inept pro- 
portions, and as far as different sizes or shapes may offer 
competing inducements, so far must they come to terms. It may 
not be a coincidence that those fish which, of all animals, change 
their proportion least in development are just those which 
grow without any known upper limit to their size. 

D''Arcy Thompson''s assay of transformations is made pretty 
well self evident by the grids superimposed upon the neigh- 
bouring figures. The somewhat arbitrary tailoring of space 
which it makes use of is fraught with metaphysical implications, 
but we must be content to observe that it has a forthright 
visual appeal, D''Arcy Thompson always compared the adult 
forms of the members of related genera or species. He com- 
pared, then, the final products of two separate processes of 
transformation, instead of comparing the two developmental 
processes themselves. Ideally he should have put both pro- 
cesses into cinematic motion, giving us two films of develop- 
ment instead of two lantern slides taken from the ends. He 
could thus have given precision to the belief that the rate of 
change of shape of animals in development, like their specific 
growth rate, progressively slows down. It is, of course, a 
generalization that is ''intuitively' obvious — a human embryo 
changes its shape more rapidly in a month than a child does in 
a year — but intuitive judgements are inoffensive only when 
everyone agrees with them, and in palaeontology, where the 
problem of the assay of form is ever with us, this is by no means 
so. But D'Arcy Thompson's method as it stands leads to the 



important inference that change of shape is orderly not only 
in time but in its spatial distribution, and that a multitude of 
particular differences of shape between two living organisms 
may be only the topical expressions of a single, simple, com- 
prehensive change of form. 

The rules of organic transformation are therefore analogous 
to those we have already arrived at in respect of growth. First, 
both size and shape change in course of development, and 
change continuously within the compass of their upper and 
lower limits. Change of size has a definite sense or trend, viz. 
of increase, and change of shape has also a definite trend. 
(Animals do indeed ''increase'* in form as they develop, if by 
that we mean that they increase in order of complexity; but 
change of complexity is outside the competence of D''Arcy 
Thompson''s method, which must confine itself to homeo- 
morphic forms.) Both then are progressive processes: it is 
exceptional for animals to grow smaller as they become older, 
and equally exceptional for them to reverse the prevailing 
trend of change of shape. Animals pass once through inter- 
mediate sizes before they reach adult weight or stature, and 
once through intermediate shapes before assuming their adult 
form. The specific growth rate is greater in early life than 
latterly, and so also is the rate of change of form. It all amounts 
to saying that growth is orderly in space as well as in its 
temporal unfolding, and that the ordinances are rather simpler 
than one might at first suppose. 

Acknowledgment is made to the Zoological Society of London 
for figure 7 (from a paper by Professor S. Zuckerman in Proceedings 
of the Zoological Society) and to the Cambridge University Press 
for Figures 8 and 9 (from D'Arcy Wentworth Thompson, On 
Growth and Form). Figure 6 is from C. Champy, Sexualite et 


The Imperfections of Man 

Evolution is one of the subjects upon which laymen have long 
felt themselves entitled to express an opinion; formerly, the 
opinion that evolution does not occur; latterly, that it does not 
occur in the way that biologists now suppose. The chief among 
several causes of their present discontent is approximately as 
j follows. Biologists believe that evolution has come about 
1 through the action of material forces, in the sense that it does 
not unfold itself according to a preordained purpose or super- 
imposed design. To laymen, an argument which takes no 
account of design or purpose or Aristotelian Final Causes is 
utterly unsatisfying and implausible. How can mere unguided 
material forces be responsible for the miraculous optical engin- 
eering of the eye; for the exquisite functional aptitude of a bird''s 
wings; for the almost finicky precision of mimicry? Is it not go- 
ing a little too far to impute these splendid accomplishments 
to what Bacon called the ^'casual felicity of particular events?' 
These are intelligible complaints, but they are founded upon 
a misconception, namely, that evolution is a perfectionist 
process. The eye, for example, is beset by chromatic and 
spherical aberration, and is not correctly centred along its 
optical axis; Helmholtz, the grand master of physiological 
optics, said that an optician would be ashamed to make an 
instrument with such elementary physical faults.* Many of the 

* [Vortrdge und Reden, I, p. 286, 1908. Helmholtz had obviously been 
exasperated by contemporary 'nature-philosophers' of the perfectionist 
school: see his Treatise on Physiological Optics, American edition, I, p. 185.] 



perplexities of laymen might be set at rest if it could be shown 
that evolution is very much a fallible, makeshift affair, and that 
loss of fitness in one regard is often the charge for some more- 
than-compensating gain. I choose the imperfections of man as 
the subject of this essay, because man''s superlative biological 
status is hardly to be questioned, and shall take three examples 
of his falls from grace: his susceptibility to haemolytic disease 
of the newborn; the mechanical shortcomings of his upright 
carriage; and the ineptitude of wound healing in injuries of his 
skin. Of these, the first is familiar enough, and I shall only deal 
with its broader aspects; the second is well understood but not 
yet widely known; and the third will be unfamiliar to all except 
a few specialists in the theory of wound healing. 

Haemolytic disease of the newborn is the general name given 
to a variety of affections (kernicterus, icterus gravis, hydrops 
fetalis) marked by grave and sometimes fatal abnormalities of 
the blood and blood-forming organs; its interpretation, which 
we particularly associate with the names of Levine, Land- 
steiner and Wiener, is one of the great triumphs of modern 
clinical biology. Briefly, it is an immuriological disease; it 
depends upon the active immunization of the mother against 
blood group substances (chiefly those of the Rhesus and Kell 
systems) absent from her own tissues, but present in the tissues 
of her child. Like many forms of allergy and hypersensitivity, 
and like the reaction that forbids the use of one person''s skin to 
repair another''s, haemolytic disease can be described as a mis- 
carriage of immunological justice — a harmful and apparently 
wanton aberration of what is properly and primarily a mech- 
anism of defence. 

Haemolytic disease of the newborn is a peculiar menace to 
human beings for the following reasons. If it is to occur at all, 
two qualifying conditions must be satisfied at the outset. First, 



the antibodies which are the chemical effectors of the immunity 
reaction must be able to pass from the mother''s circulation into 
the circulation of the unborn child. In effect, this means that 
the membranes which separate mother from foetus must be of 
such a kind as to let the antibodies through. This first condi- 
tion is satisfied by rabbits, rats and mice, and also by human 
beings, but not by cattle, sheep and horses. Second, the foetus 
must reach before birth a stage of development at which the 
immunizing substances present in its blood corpuscles have 
reached maturity, for if they are still undeveloped at birth the 
mother can have no normal opportunity to become immunized 
against them; and even if the mother were to be artificially 
immunized, the antibodies so formed could not attack the 
foetal blood. This second condition is satisfied by cattle, sheep 
and horses, and also by human beings, but not by mice and 
probably not by rats. 

On the face of it, mice should be specially liable to be 
immunized by their own young, for a female mouse may give 
birth to a quarter of her own body-weight of young in a single 
pregnancy and to ten times her body-weight in a lifetime; 
furthermore, antibodies formed in the mother have ready 
access to the embryos within it. But it seems to be impossible to 
give mice haemolytic disease, even when their mothers are 
deliberately immunized against the red blood corpuscles of 
their young. Mitchison attributes this to their extreme im- 
maturity at birth; if the degree of maturity of red blood cells is 
taken as a yardstick, birth in mice takes place at a stage equi- 
valent to a human embryo that is still six months from term. 
Conversely, cattle, horses and sheep can only get haemolytic 
disease after birth — a stage at which it is clinically manageable 
— for it is only at birth that antibodies pass from the mother to 
her young, in the colostrum, the first watery milk. 

Human beings qualify on both counts: maturity at birth and 
the ability of antibodies to gain access to the unborn young. 



Nor is this the whole story. Haemolytic disease can only occur 
if the members of an interbreeding population are dissimilar in 
their antigenic make-up. If all human beings were Rhesus- 
positive or Rhesus-negative, it is obvious that they could not 
get haemolytic disease (or suffer from transfusion accidents) as 
a result of immunization by the Rh antigens. But they are, as 
it happens, most highly diverse with regard to the antigens 
present in their red blood corpuscles. About one in six English- 
men lack the most mischief-making antigen of the Rh series, 
viz. D or Rhg; about one in ten possesses the Kell antigen. 

We may now ask, what advantage do human beings enjoy 
which compensates, or more than compensates, for their vul- 
nerability to haemolytic disease? It is clear that man's embryo- 
logical advantages, if such they are — a long gestation period, 
coupled with a form of gestation which allows beneficial as well 
as harmful antibodies to enter the foetus — could be enjoyed 
with impunity if the entire population were either all positive 
or all negative in respect of the blood group substances Rh 
and Kell. The question therefore resolves itself into asking: 
Why, then, have they not become so? Unfortunately, the 
answer is not known; it is merely being groped after. 

Roughly speaking, there are two kinds of reasons why a 
population should be polymorphic, i.e. should be subdivided 
into variant types of which even the least frequent is far too 
frequent to have originated merely through the recurrence of 
mutations. The first is that heterozygotes (which carry and 
therefore propagate a gene, but do not necessarily reveal its 
presence) should stand at some special advantage relative to 
homozygotes. Contrary to all superficial appearances, this 
appears to be true of the blood affection known as 'sickle-cell 
trait", the analysis of which provides a most noteworthy 
example of a combined operation in genetics, chemistry, 
anthropology and clinical medicine. 

Briefly, 'sickle-celP trait is an affection in which the red cells 



of deoxvgenated blood adopt a sickle shape. Pauling and his 
colleagues showed that it is due to the presence of an abnormal 
variant of adult (as opposed to foetal or infantile) haemoglobin. 
Sickle-cell trait appears in about 9 per cent of American 
negroes, and in a proportion varying from to 45 per cent in 
communities of African negroes. Its inheritance, worked out 
independently by Neel and Beet, is governed by a Mendelian 
dominant gene, and sickle-cell trait (a condition which is not 
harmful in itself, nor even appreciably disabling) is its hereto- 
zygous or hybrid form of expression. Individuals who are 
homozygous for the sickle gene, however, suffer from a grave 
and sometimes fatal anaemia, and few of its sufferers live to 

In the face of powerful selection against the homozygous 
form the frequency of the sickling gene is far greater than can 
be accounted for by mutation. According to Allison, the reason 
why it flourishes is that the heterozygote is endowed with a 
specially high resistance to subtertian malaria; in areas of the 
world where malaria is hyperendemic the gene is therefore 
kept in being by the high selective advantage of the hetero- 
zygote; elsewhere it is going or gone. 

No such neat and rounded story can be told of blood group 
polymorphism; indeed, it was at one time thought (and by 
geneticists feared) that the subdivision of human beings into 
the blood groups A, B, AB, O was entirely capricious, in the 
sense that an individual's blood group had no bearing on his 
fitness to survive and reproduce. Workers at the British Post- 
graduate Medical School now find that membership of blood 
group A is associated with an increased susceptibility to cancer 
of the stomach, and membership of group O with a greatly 
increased susceptibility to peptic ulceration. Blood group poly- 
morphism is thus certainly not a matter of indifference, though 
its import is still obscure. But, so far as I am aware, no one has 
yet been able to associate the subdivision into Rh blood types 



with anything except the unquahfied incubus of transfusion 
accidents and haemolytic disease. 

It is possible, then, as Haldane has suggested, that the 
diversity of Rh blood types represents a second kind of poly- 
morphism — that which is merely transient, a necessary inter- 
mediate stage betv/een the elimination of one type or the other. 
This interpretation is borne out by the existence of rather bold 
inequalities among different races and nations in the propor- 
tions which belong to the several groups. The number of 
Basques who are Rh-negative falls between one in three and 
one in four, but Levine and Wong found only one Rh-negative 
individual among 150 Chinese, having been led to their 
enquiry by the significant observation that haemolytic disease 
in China is very rare. If the interpretation is true, then haemo- 
lytic disease could be explained away as a transient genetic 
ailment of mankind, but fortunately we can look forward to 
something a little more expeditious than an evolutionary cure. 

Man''s upright carriage may be a constant source of moral 
satisfaction, but it has certain serious mechanical drawbacks. 
Man is unique among four-legged animals in being able to 
stand erect, on the flat of his feet, and to balance himself in 
that position. (Even gorillas do not stand upright more than 
momentarily, and they walk not on the flats but upon the 
outer margins of their feet.) The shape of the backbone has 
changed accordingly. In all other animals, with unimportant 
exceptions, the backbone is more nearly horizontal than vertical, 
and it takes the form of a single unkinked or uninflected arc 
from neck to tail. The ''vertebral column"* is not a column at 
all, but is more like a cantilever having the four legs as piers. 
The vertebral column of a human being is no longer a simple 
uninflected arc; it bends slightly forwards in the neck, slightly 
backwards in the thoracic cage; forwards again in the lumbar 



region, the small of the back, and backwards in the fused 
vertebrae that form the sacrum. That is the mature pattern; in 
development, the neck flexure appears somewhat before birth, 
and the lumbar flexure between the ninth and eighteenth 
months of age. 

An upright stance imposes new and peculiar stresses upon 
the spinal column. The support of weight imposes a force acting 
down the vertical long axis, which tends to compress the 
vertebrae upon themselves. The angle of their apposition is 
responsible for a shearing force between the bottom-most 
lumbar vertebra and the sacrum; and general flexional strains 
become very apparent when stooping to pick up weights. To 
cope with these new forces (for such, in an evolutionary sense, 
they are) man inherits only the standard outfit of muscles and 
ligaments, and the muscular bracing of the neck and lumbar 
region leaves much to be desired. 

What suff'ers from the w^ear and tear of habitual use is not, 
primarily, the vertebrae themselves, but the tissues lying 
between them. The bodies of the vertebrae are not set against 
each other face to face; on the contrary, about one-quarter of 
the total height of the column (more in the lumbar region) is 
occupied by peculiar solid intervertebral joints. Each joint 
forms a so-called intervertebral disk — a central nucleus of semi- 
fluid consistency, which embodies or represents the remnant of 
the embryonic notochord; contained within a tough fibrous ring, 
the annulus, in which the fibres are disposed cylindrically in 
coaxial rings; the whole being bounded above and below by 
flat cartilaginous plates. The whole organ has been described 
by one of its leading students, Ormond Beadle, as a hydro- 
dynamic ball bearing. 

The bearings may give in a variety of difl'erent ways. Under 
repeated flexion of the spinal column when it is taking weight, 
the nucleus may gape through a weakness in the fibrous ring 
which normally contains it, press against the posterior liga- 



ment, and even encroach upon the spinal canal. Alternatively, 
as if by the insistent action of "'telescoping'' forces, nuclear 
matter may break through perforations in the cartilaginous 
plates and obtrude into the vertebral bodies, which are made 
of spongy rather than of compact and concentrated bone. In 
recent years, the opinion has been gaining ground (perhaps too 
rapidly) that many disabilities which have been loosely classi- 
fied as sciatica, lumbago and vague rheumaticky back pains are 
due to abnormalities of the intervertebral disks; the immediate 
causes of pain may still be debatable — for example, as to 
whether or not mere chronic pressure on a nerve root can cause 
inflammation and thereby pain — but their anatomical origin 
seems pretty certain. 

The exhaustive anatomical studies of the Dresden patholo- 
gist Schmorl have led the way to a conclusion of more 
general biological interest, that the spinal column is the first 
organ in man to ''age\ that is, to show the deterioration conse- 
quent upon ageing.^ Pathological changes, it has been said, are 
detectable as early as the eighteenth year of life. The deteriora- 
tion of the spinal column provides, indeed, what is perhaps the 
best example of a process of ageing which at least begins by 
being a consequence of the cumulative effects of wear and tear 
— of chronic functional attrition, as opposed to the ""innate 
deterioration**, which takes place independently of abuse, or 
even use. 

Man''s upright stance has incomparable advantages, perhaps 
above all in providing for his principal physical (as opposed to 
mental) asset, manual dexterity, but it takes its toll in the 
mechanical vulnerability of the spine. 'Disk lesions'* are not, of 

1 It has competitors of course. If we accept the widely held view that the 
human ovary starts with a fixed number of egg cells, which cannot be added 
to but are merely used up as life goes on, then the ovary could be said, in a 
sense, to 'age' from birth. But it is a rather special sense. A clock may work 
perfectly though its mainspring is uncoiling; it is using up its potential of 
stored time, but as a timepiece it is as good as fully wound until it stops. 

I 129 


course, peculiar to man, but it does not derogate from our 
argument that they should also occur in dogs, for they are 
found principally in those breeds (bull-dogs, pekinese) which 
have been deliberately selected for imperfect cartilaginous 
development. Only we ourselves, therefore, not natural evolu- 
tion, can be held to blame. 

The third of man's peculiar shortcomings on our agenda 
is the appalling ineptitude of wound healing in the skin, and 
here I shall follow closely the reasoning of Billingham and 
myself. The following comparison will make the problem 

If the entire thickness of the integument in the chest region 
of an adult rabbit is excised over a rectangular area of 100 cm.^; 
something that looks superficially like an irreparable injury is 
produced. But, so far from being irreparable, it requires for its 
quick and successful healing nothing more than the most 
elementary surgical care. The surface area of an adult human 
being is about seven or eight times as great as a rabbifs, but a 
skin defect of the same absolute size and depth, and the same 
relative position, cannot by any means be reHed upon to heal 
satisfactorily of its own accord. If left to itself, it will heal pain- 
fully slowly, and will gather up and scar; a wound of similar 
size in the leg (which is not so much thinner than a rabbit's 
trunk) could cause a seriously disabling injury if left untreated, 
whether by gathering up in such a way as to constrict the blood 
supply of the Umb or by immobilizing a joint. Such an injury 
cries aloud for skin grafting, an operation in which a thin flat 
slice of normal skin is removed from some undamaged part 
of the patient's body and held in place for four or five 
days over the area of loss. (The skin graft is removed so 
thinly as to leave behind part of the leathery layer of the 
skin of the donor area, and the bases of the hair shafts; the 



donor area will therefore heal of its own accord without scar- 
ring or contraction.) 

The question is, why is the rabbit so accomplished in wound 
healing and the human being so strikingly inept; the answer 
turns upon an understanding of the mechanism of healing as 
it occurs in a rabbifs skin. 

As Billingham and I see it, there are two quite distinct 
(though concurrent) elements in the healing of rabbits'* skin. 
The first is contracture, which closes the wound by a progres- 
sive coming together of its original edges. Contracture follows 
a regular geometric pattern: starting as a rectangle, the wound 
first becomes smaller wdthout changing shape; then the sides 
cave in towards the centre, and meet from the four corners 
inwards, so that all that is left is a neat )> — <^ shaped line of 
suture (it does not deserve to be called a scar). During the 
process of contraction the raw wound area is temporarily 
covered by a thin film of skin epithelial cells which grow or 
migrate inwards from the edges of the wound; but as the 
original skin edges come together, so the space enclosed by 
them diminishes and finally disappears, and the skin epithelium, 
which is purely a temporary organ of healing, disappears 
with it. 

Contracture closes the wound, but it does not, of course, 
make good the loss of 100 cm.^ of skin. Billingham and I believe 
that this loss is made good by a second process, the intercalary 
or ''intussusceptive^ growth of the remaining skin. Intercalary 
growth is an expansion of the skin by growth on or within its 
existing fibrous framework. The simplest way of demonstrat- 
ing it is to prepare a rectangular wound and to leave behind, 
in its centre, a small area of undamaged skin. Alternatively, 
skin may be excised from the whole of the area and a skin graft 
thereupon placed in the middle. The forces of contracture 
which draw the skin edges together bring an expansive force to 
bear on the central island, which may accordingly enlarge to 



no less than ten times its original area. The number of hair 
roots is not added to,^ so that the most obvious outward 
evidence of intercalary growth is the fact that the hairs growing 
from the central island become spaced apart to a degree pro- 
portional to its linear enlargement. There is nothing particu- 
larly novel or mysterious about the intercalary enlargement of 
skin, for it is a process that occurs naturally in growth from 
newborn to adult size. It is made particularly obvious by the 
fact that when a child is grafted with skin, the graft grows 
with him. 

Contracture and intercalary expansion between them make 
for admirably efficient healing in rabbits. Why do they not 
work to equally good effect in men? 

The answer is probably anatomical. Contracture can only 
be an efficient healing process if the skin is sufficiently loose to 
'■give'' while it is going on. In rabbits, and in mammals generally, 
the integument is very loosely knit to the body wall; its main 
blood supply runs in a plane parallel to the surface, and it 
contains its own intrinsic musculature (the ""panniculus 
carnosus"") which makes it possible for mammals to twitch their 
skins. In human beings, the integument is no longer a gener- 
ously fitting coat, but is much more firmly knit to the tissues 
below; the intrinsic muscles of the skin are now confined to 
areas of the face and neck, and the skin generally is much more 
of a piece with the rest of the body. The upshot of this new 
anatomical arrangement is that contracture, so far from being 
an efficient mechanism of wound closure, has become something 
of a menace; it constricts, disfigures and distorts, and may yet 
fail to bring the edges of the wound together. But it still occurs: 

^ Dr Breedis has recently shown tliat, contrary to almost all expert belief, 
the number of hair roots can be added to in adult animals, and my col- 
leagues R. E. Billingham and Paul Russell have done experiments which 
entirely bear out his views. The new formation of hair roots, however, takes 
place only under certain special circumstances that do not affect experi- 
ments of the kind described above. 



human beings may be said to have retained the mechanism of 
heahng by contracture, but to have lost the anatomical pre- 
requisites which enable it to proceed to good effect. As a 
mechanism of wound healing, the contracture of human skin 
is therefore as archaic as the vermiform appendix; like the 
appendix, we become aware of it only when it leads to harm. 
Fortunately, thanks to the ingenious and entirely artificial act 
of skin grafting, human beings need no longer suffer the dilatory 
and incompetent ministrations of the ""naturaP process of 
repair. What compensating advantage the human being gets 
from the novel structure of his skin is far from obvious, though 
it is hard to believe that there is none. 

What I have sought to show in this article is that evolution- 
ary advancement is a compromise between what is desirable 
in the abstract and what can in fact be done; that the lesser 
evil must be put up with if it makes possible the greater good; 
and that bad mistakes are made which, though foreseeable to 
a prescient mind, were not in fact foreseen. The philosophic 
import of this proposition may well be most debatable, but the 
truth of the proposition itself can hardly be in doubt. 


Tradition : 
The Evidence of Biology^ 

In order to avoid any possible misunderstanding, I want to 
make it clear from the beginning that I am going to address you 
as a professional biologist, and that I shall consider only that 
fraction of human behaviour about which a biologist might be 
expected to have something pertinent to say. I shall touch very 
briefly upon two problems. First, what is to be learned about 
the causes and motives of human behaviour — about our Springs 
of Action — by thinking of man as 'just another animal', that 
is, by thinking of the biological similarities between animals 
and men? Everybody recognizes that there are indeed profound 
similarities between the behaviour of man and animals, but 
biologists and laymen think about them in entirely different 
ways. When laymen see mice nursing and cherishing their 
young, their first thought is ''How like human beings they are, 
after all!' The biologist (at all events when he is on duty) thinks 
*'How mouse-like, after all, are men!' 

The second question is, what is to be learned by reflection 
upon the biological differences between men and other animals.'^ 
In answering this question I shall come to a conclusion that 
may surprise you, viz. that tradition is responsible for a large 
part of the present biological fitness of man. 

^ [The gist of one of several short addresses on 'Tradition' given at a 
Present Question Conference in 1953 on the general theme of Springs of 


tradition: the evidence of biology 

In everyday life (as opposed to conferences) we never think 
about Springs of Action in a general way at all; we think only 
about the springs o^ particular actions and with the problem of 
choosing between one action and another. We do not worry 
about why human beings have propensities for loving and 
hating, but about why one person loves a second and hates a 
third. We take it for granted that people need food and take 
steps to get it, but what is interesting about food-seeking 
activities of human beings is why they eat this and not that, 
here and not there, now and not then. It is no great new truth 
that human beings are ambitious; what is interesting about 
ambition is why in one person it takes the form of wanting to 
become a great musician, in another of wanting to raise a large 
family, and in a third (for this too is an ambition) of wanting 
to do nothing at all. In these three examples I hope you can 
see a clear distinction between the propensities underlying 
certain general kinds of behaviour and the factors which decide 
that a certain general kind of behaviour shall take a certain 
particular form. 

Unfortunately, the evidence of biology does not yet run to 
analysing the sources of particular human actions and decisions: 
that is a matter for psychology, or perhaps for common sense. 
But that does not mean that the evidence of biology is un- 
informative or dull. Tinbergen^ and Lorenz have given us 
reasons for believing that many kinds of behaviour which seem 
to us to be peculiarly human are part of a very ancient heritage 
— 'showing off, for example; playing with dolls; sexual rivalry; 
and many kinds of '"displacement activity\ in which a thwarted 
instinctive impulse vents itself in actions of an apparently quite 

1 See N. Tinbergen's The Study of Instinct, Oxford, 1951. My indebted- 
ness to Tinbergen will be very obvious to anyone who follows the newer 
research on animal behaviour ('ethology', as it has come to be caUed). 



irrelevant kind. In the main, though, the evidence of biology 
serves only to identify the parameters in the equations of 
human behaviour, if '"instincts' can be so described. 

It is not at all easy to define instinctive behaviour, but it 
has certain properties that distinguish it clearly from be- 
haviour of other kinds. First, instinctive behaviour is unlearried. 
In practice, it is sometimes very difficult to decide whether a 
particular act of behaviour is learned or unlearned, and with 
human beings it may be well-nigh impossible. With animals it 
is simple enough in principle. Is nest-building activity in mice 
and birds copied or inborn? To answer this question decisively, 
one must rear a mouse or bird in strict isolation from birth, 
away from any possible source of information about how nests 
are made; and one finds that a mouse so reared can make itself 
an admirable nest. This provides one good reason for describing 
nest-building as an instinctive act. Then again, instinctive 
activities are '"purposeful". Never mind the teleological under- 
tones of the idea of purpose; all I mean is that instinctive 
activities, however complex, do in fact converge upon a certain 
goal — mating, feeding, or whatever the case may be. The 
functional import of instinctive activity, and the pattern of the 
connections between its several parts, is intelligible only by 
reference to the goal towards which it is directed. Thirdly, 
instinctive behaviour has a Mrive"*, a sort of psychological 
pressure behind it; a drive, for example, to find food or to find 
a mate. The drive is temporarily discharged or assuaged by the 
act which constitutes the goal of a particular instinctive action. 
It is usually possible to distinguish two phases in a sequence ol* 
instinctive behaviour: first, ''appetitive behaviour'', that earlier 
phase in which an animal seeks the means of gratifying its 
instincts; and second, the performance of the ''consummatory 
act** which finally achieves the goal. The appetitive behaviour 
that gives evidence of a hunger drive includes all the activities 
entailed by seeking food. The consummatory act of the instinct 


tradition: the evidence of biology 

is eating, and \vith its performance the hunger drive is dis- 
charged or worked off, and that particular episode of in- 
stinctive activity comes to an end. 

Students of animal behaviour have described, analysed and 
then pieced together again a great variety of different kinds of 
instinctive action. Two conclusions which can be drawn from 
their work, though both are negative, have a profound bearing 
on human affairs. There is no such thing as an ''aggressive 
instinct"*, and it is therefore altogether wrong to suppose that 
human beings can be its victims or its beneficiaries. There is 
no drive, no motive force in animal behaviour that is dis- 
charged or gratified by the mere act of fighting. Fighting and 
aggression — much of it bluff — do indeed play a part in animal 
life, but they are entirely subsidiary or incidental to certain 
other complex instincts. Males may fight in establishing their 
mating territories, and fighting may play a part in seeking food 
or in defence, but there is no such thing as an 'aggressive 
instinct' in itself. There is simply an aggressive element in 
several instincts, as there may well be an element of co-opera- 
tion or mutual aid. There is equally no such thing as a 'social 
instinct% no sort of inward compulsion that is set at rest 
merely by getting together in groups; but co-operation is, as 
fighting is, a component of several different kinds of instinctive 

Let me say just one other thing about the role of instinctive 
activities in human conduct, using this quotation from Alfred 
North Whitehead as a text: 

It is a profoundly erroneous truism, repeated by all copy 
books, and by eminent people when they are making speeches, 
that we should cultivate the habit of thinking what we are 
doing. The precise opposite is the case. Civilization advances 
by extending the number of important operations which we 
can perform without thinking about them. 

This is a most important and arresting half-truth — so com- 



pellingly true, or half true, that one wonders how anyone could 
ever have held a contrary opinion. But it is true only oi learned 
activity. No matter what the activity may be — learning the 
multiplication table, or how to drive a car, to speak intelligibly, 
or to sew — learning is a process of thinking and deliberation 
and trial and decision, but the state of having learned is the 
state in which one need think no longer. Paradoxically enough, 
learning is learning not to think about operations that once 
needed to be thought about; we do in a sense strive to make 
learning ""instinctive"*, i.e. to give learned behaviour the readi- 
ness and aptness and accomplishment which are characteristic 
of instinctive behaviour. But that is only half the story. The 
other half of the half truth is that civilization also advances by 
a process which is the very converse of that which Whitehead 
described: by learning to think about, adjust, subdue and 
redirect activities which are thoughtless to begin with because 
they are instinctive. Civilization also advances by bringing 
instinctive activities within the domain of rational thought, 
by making them reasonable, proper and co-operative. Learn- 
ing, therefore, is a twofold process: we learn to make the 
processes of deliberate thought 'instinctive"* and automatic, 
and we learn to make automatic and instinctive processes 
the subject of discriminating thought. 


I now want to try to answer the second question I put before 
you: in what fundamental biological way do human beings 
differ from other animals? One possible answer, which I shall 
try to justify, is this: man is unique among animals because of 
the tremendous weight that tradition has come to have in 
providing for the continuity, from generation to generation, of 
the properties to which he owes his biological fitness. 

It is the merest truism that man is a tool- or instrument- 


tradition: the evidence of biology 

using animal. The instruments used by human beings are of 
two chief kinds. The first I shall call motor or effector instru- 
ments — for example, hammers, cutlery, motor-cars, mega- 
phones and guns, instruments which increase, sometimes 
prodigiously, our repertoire of motor activity. Instruments of 
the second kind can be described as sensory accessories: 
spectacles, ear-trumpets, radio sets, thermometers, appliances 
which increase beyond all former bounds the competence of 
one''s ordinary senses. (Not all instruments fall into these two 
categories: clothes, for example, do not.) Man is not quite 
uniquely an instrument-using animal; but the odd examples of 
the use of tools by lower animals are so rare that each one is 
treasured and made a fuss of. The Galapagos woodpecker, a 
sort of finch, uses a thorn held in its beak to prise insects from 
the bark of trees. Many animals make houses or shelters, 
but these are tools of a kind I shall not be concerned with 

I propose to use the terms invented by the great actuary 
A. J. Lotka to distinguish between the organs that we are born 
with and organs that are made: e?idosomatic instruments for 
eyes, claws, wings, teeth and kidneys, exosomatic instruments 
for telescopes, toothpicks, scalpels, balances and clothes. 
Although there is a very obvious distinction between instru- 
ments of these two kinds, the distinction is much less obvious 
biologically than it is to the unaided power of common sense. 
The two kinds of instrument serve the same biological func- 
tions, and each can to some extent deputize for the other. Even 
in a quite narrowly biological sense, man is a flying animal: he 
can fly faster, farther and higher than birds, if not yet with 
quite the same finesse. It is also important to remember that 
exosomatic instruments Sive functionally parts of the body, even 
if they are anatomically separate and distinct. All sensory tools 
like spectacles, Geiger counters and spectrophotometers report 
back at some stage and by some route through the ordinary 



endosomatic senses, and motor instruments are functional, and 
functionally intelligible, only when they are used. It is not 
spectacles, but spectacles worn and looked through, that are 
instruments of vision, and the hammer is only a tool when it is 
wielded by the hand. (I think it was Wilfred Trotter who said 
that when a surgeon uses a simple instrument like a probe or 
seeker, which is merely an extension of the fingers as stilts are 
extensions of legs, he actually refers the sense of touch to its 
tip.) The relationship between instrument and user may be 
very remote, as it is with guided missiles and with engines 
designed to work without attention, but their conduct is built 
into them by human design and, in principle, their functional 
integration with the user is just the same. It is for this reason 
I deplore the habit of describing the brain as a kind of calcul- 
ating machine; the truth is that a calculating machine is a kind 
of exosomatic brain. It performs brain-like functions, much as 
cameras have eye-like and clothes have skin-like functions, and 
motor-cars the functions endosomatically performed by legs. 
We may indeed learn something about the brain by studying 
calculating machines, as we have learned something about the 
eye by studying lenses; but it need not be so: the internal- 
combustion engine has no lessons to teach us about how 
muscles work. 

Biologists in the nineteenth century were much impressed by 
the fact that exosomatic instruments undergo a systematic 
secular change that is somewhat analogous to ordinary bio- 
logical evolution. Just as legs and ears have changed in the 
course of time, so also have bicycles, microscopes, radio sets 
and cars. The evolution of both endosomatic and exosomatic 
organs is gradual in synoptic view, but somewhat discontinuous 
on closer inspection. Novelties arise in both, not in the entire 
population but in a limited number of its members, and may or 
may not spread thereafter through the whole. Both modes of 
evolution are "integrative"' in the sense that they start or pro- 


tradition: the evidence of biology 

ceed from the groundwork of what has been achieved before. 
Failure, abortion or extinction occur as commonly in the one 
as in the other; 'vestigial organs'*, like the coccyx or appendix, 
or those functionless buttons on the cuffs of men"'s coats, occur 
in both. Whether or not you choose to describe the systematic 
secular change of exosomatic instruments as an ''evolution'' is 
utterly unimportant; all I am concerned to emphasize is that 
both exosomatic and endosomatic ''evolution'' are equally 
modes of the activity of living things, and that both are 
agencies — to some extent alternative agencies — for increasing 
biological fitness, i.e. for increasing those endowments which 
enable organisms to sustain themselves in and prevail over 
their environments. In man, ordinary evolution as we under- 
stand it in lower animals, endosomatic evolution, does still 
happen, and I could give examples of evolutionary changes 
that have occurred within the known history of the human 
race. But they are changes of a comparatively minor character, 
whereas the changes wrought upon human society by exo- 
somatic evolution have been rapid and profound. 

I now at last come to the point. There is one crucial dis- 
tinction between endosomatic and exosomatic evolution. 
Ordinary evolution is mediated by the process of heredity. 
Exosomatic ''evolution'' (we can still call it ''systematic secular 
change'') is mediated not by heredity but by traditio7i, by 
which I mean the transfer of information through non-genetic 
channels from one generation to the next. So here is a funda- 
mental distinction between the Springs of Action in mice and 
men. Mice have no traditions — or at most verv few, and of a 
kind that would not interest you. Mice can be propagated from 
generation to generation, with no loss or alteration of their 
mouse-like ways, by individuals which have been isolated from 
their parental generation from the moment of their birth. But 
the entire structure of human society as we know it would be 
destroyed in a single generation if anything of the kind were 



to be done with man. Tradition is, in the narrowest technical 
sense, a biological instrument by means of which human beings 
conserve, propagate and enlarge upon those properties to 
which they owe their present biological fitness and their hope 
of becoming fitter still. 



The Uniqueness 
of the Individual 


Philosophy and common sense, though often parted, have long 
agreed about the uniqueness of individual man. Different men 
have different faces, sizes, shapes and origins; different apti- 
tudes, skills and predilections; and different ambitions, hopes 
and fears. Science now makes it a trio of concordant voices, for 
the uniqueness of individual mice and men is a proposition 
which science can demonstrate with equal force, perhaps with 
deeper cogency, and certainly with a hundred times as much 
precision. For reasons that will not become apparent until 
later, I shall begin what I have to say wdth some observations 
on the repair of burns. 

Deep and extensive burns are injuries which cannot heal 
properly of their own accord. What happens when a severe 
burn is left to heal by its own devices is, in outline, this. The 
skin which has been killed by burning eventually sloughs away; 
its place is taken by a spongy, moist, highly vascular tissue of 
repair called granulation tissue. Repair, such as it is, is brought 
about by two concurrent processes. The epithelial cells which 
form the outermost layer of the skin creep inwards over the 
surface of the granulation tissue, and as they do so, the granu- 
lation tissue begins to form connective tissue fibres similar in 
individual make-up to those which form the leathery layer of 



the skin but arranged in a different and functionally ineffective 
pattern. In the meantime, the edges of the wound are forcibly 
drawn together by the tensile forces generated within the 
granulation tissue: this is the process of contracture. In rabbits 
and other mammals with a loose integument, contracture is the 
normal mechanism of repair, and it works admirably, for a 
rabbifs skin is so mobile that the edges of the wound can be 
drawn together into the neatest possible natural line of suture. 
Human beings retain the mechanism of healing by contracture, 
but human skin is so firmly bound to the tissues underneath it 
that contracture is no longer an efficient method of repair; the 
edges of the wound are forcibly dragged inwards, instead of 
giving easily, and the skin around is gathered up and distorted. 
(Mere disfigurement is one of the lesser evils, for in certain 
parts of the body contracture can constrict the blood-vessels, 
acting like a tourniquet, or immobilize a joint.) The end-result 
of this entirely inept process of repair can only be described as 
functionally and cosmetically abominable. A greater or lesser 
scar is left, made of dense fibrous tissue, and covered by a 
sometimes unstable and always unsightly surface layer of 
epithelium which never regains its natural suppleness and 
colour nor grows anew its normal endowment of glands and 
hairs. But before the surgical innovations I am about to 
describe came into common use, to achieve even such an end- 
result as this would be a matter for congratulation, for ''naturaP 
repair is a dilatory process that gives the patient every reason- 
able opportunity to die from the steady seeping away of body 
fluids through the wound's raw surface, or from the wound 
infections that, without the help of antibiotics, an already 
enfeebled body could do little to oppose. 

For this appalling problem, the surgical operation of skin 
grafting provides an almost complete solution. What is done, 
as a rule, is this. A broad, thin sheet of skin is removed from 
some uninjured part of the patient"'s body, most conveniently 



the thigh, laid upon the area from which skin has been lost, 
and held firmly in position for four or five days until the 
primary union of the graft to the tissue underneath it is com- 
plete. The area from which the graft was removed will heal of 
its own accord within a week or two, for the graft will not have 
been cut so thick that epithelium from the inner ends of the 
truncated hair roots cannot creep upwards and grow over the 
denuded surface; indeed, one donor area can provide more than 
one crop of skin. 

The use of grafting to make good the loss of skin is satis- 
factory for wounds up to, perhaps, one square foot in area; 
but (in the form in which I have described it) it becomes less 
and less satisfactory as the area of the wound increases, and 
a severely burnt patient may well have lost far more skin than 
can be wholly made good from his own resources. The area of 
loss cannot now be fully covered with grafts; what is done, 
therefore, is to ''seed'' it with small patches of skin in the shape 
of squares or rectangles, evenly spaced apart. The outgrowth of 
epithelium from these little skin grafts, combined with in- 
growth from the wound margins, forms a new skin surface, and 
contracture, though it still happens, is reduced in proportion 
to the area that has been covered with grafted skin. This 
operation of patch-grafting is avowedly a makeshift; the end- 
result is neither elegant nor functionally more than serviceable; 
but it is designed to save a life which might otherwise have 
been despaired of, and so it does. 

What is interesting to the biologist, however, is not what the 
surgeon does in such a predicament but what he does not do. 
If a patient cannot afford skin of his own for grafting, why not 
use skin grafts from someone else? There is no surgical obstacle 
to such a procedure, and voluntary donors are not hard to 
come by; nor would there be any difficulty in setting up a 
'skin bank** to be drawn upon in an emergency, for skin may 
be stored alive and without deterioration for a good six months 

K 145 


at the temperature of liquid air or carbon dioxide snow. The 
difficulty is that a skin graft from one human being will not 
form a permanent graft upon the body of another. In the first 
week or so after its transplantation the homograft (as it is 
called) behaves just like a graft which has merely been trans- 
posed from one part to another of a single individual. The first 
outward sign that all is not well is a puffiness and inflammation 
of the grafted skin, leading to a weakening and ulceration of 
its surface, and finally to abject necrosis followed by a slough- 
ing away. How soon this happens — perhaps after only ten days, 
perhaps in a month — depends upon a number of variables, 
some of which I shall mention below. Every now and again a 
homograft lasts long enough to make a surgeon begin to hope 
that a natural law is about to be suspended in his favour, but 
sooner or later (and the general rule is sooner rather than later) 
the graft withers up and disappears. A human being is resol- 
utely intolerant of skin grafted upon him from other members 
of his own species; so is a newt, chicken, mouse or cow; nor will 
even a goldfish accept a scale from any other.^ The problem 
of how this comes about, why it should be so, and what can be 
done about it is the subject of the present article. 


The idea that homografts of skin are invariably destroyed 
within a few weeks of their transplantation, so that they are 
useless for anything except avowedly temporary repair, is now 
accepted by all well-informed surgeons, though recognition of 
the truth was slowly and hardly won. I now wish to consider 
three exceptions to this general rule, two of them predictable 
and perfectly intelligible, the third of a surprising and entirely 
unexpected kind. 

The first exception is that skin and other tissues can be 

1 W. H. Hildemann, Proc. N. Y. Acad. Sci., 1957. 



exchanged between identical twins on a scale limited only by 
the exigencies of surgical technique. For the purpose of trans- 
plantation, identical twins behave as if they were a single 
individual — as indeed they once were, for both arose from a 
single fertilized egg. Skin grafting has just once been used to 
clear up a problem of disputed parentage.^ One day the father 
of two six-year-old boys Pierre and Victor (HI s^agit ici de 
pseudonymes'') had his attention called to a third small boy who 
was reported to be the very image of his Victor. The third boy, 
Eric, had been born in the same clinic and on the same night; 
it seemed possible, therefore that Eric and Victor were twins 
and that the children had somehow been muddled up. A very 
careful character-comparison undertaken by Professor Frances- 
chetti made it all but perfectly certain that Eric and Victor 
were indeed identical twins, and at least likely, therefore, that 
Pierre really belonged to Eric'*s putative mother; but this was 
an hypothesis which she was unwilling to entertain. Blood 
grouping tests of some refinement now revealed that Pierre 
could not have been the son of his alleged mother, but without, 
unfortunately, excluding the possibility that Eric was his 
alleged mother's son. (Eric'*s father was dead.) In this predica- 
ment. Sir Archibald Mclndoe was asked to perform a test by 
which the case would be agreed to stand or fall: small skin 
grafts were exchanged on the one hand between Pierre and 
Victor and on the other hand between Victor and Eric. The 
skin grafts exchanged between Pierre and Victor were de- 
stroyed and sloughed away in a matter of weeks; those ex- 
changed between Victor and Eric survived over the whole 
period of observation and presumably survive still. These tests 
can be taken as a proof that Victor and Eric were twins, and 
under the circumstances, a proof of 99*9 per cent certainty that 

1 A Franceschetti, F. Bamatter and D. Klein, Bull. Acad. Suisse Sci. 
Med., 4, p. 433, 1948; A. Mclndoe and A. Franceschetti, Brit. J. plastic 
Surg., 2, p. 283, 1950. 



they were identical twins. (The reason for this pawky reserva- 
tion will be explained later.) As to Pierre and Victor, the skin 
grafting test proved nothing except that they were not 
identical twins; but as blood grouping had shown that Pierre 
had certainly been allotted to the wrong mother, Eric'*s mother 
at last accepted him as her own. 

Just as skin grafts can be exchanged between identical 
twins, so also can they be exchanged between mice and guinea- 
pigs which, by having been inbred brother to sister for upwards 
of twenty successive generations, have come to resemble each 
other in hereditary make-up almost as closely as if they were 
identical twins. (This is not necessarily true for all animals, for 
some have mechanisms which prevent their achieving a very 
high degree of genetic uniformity.) But even here there is a 
curious exception of some theoretical importance^: in some 
inbred strains of mice, females will not permanently accept 
skin homografts from males. It is now pretty well certain that 
this is because agents of the type that cause the breakdown 
of homografts are present on, or caused to be formed by, the 
Y-chromosome, that which is peculiar to males. There is 
accordingly no reason why females should not accept grafts 
from other females, nor males from females or other males, 
as indeed they do. 

When members of two different inbred lines of mice are 
crossed (supposing them already to have achieved and retained 
a sufficiently high degree of genetic uniformity), their hybrid 
progeny, forming the so-called F^ generation, are also uniform, 
and will accept grafts from one another. They will also accept 
grafts from their parents and, mutations apart, from their own 
progeny of the first or any subsequent generation. Excepting 
only the special case of grafts transplanted from males to 
females, F^ hybrids are therefore the ""universal recipients'* of a 
little microcosm of mice comprising the inbred parental strains 

1 E. J.Eichwald and C.R.Silmser, Transplantation Bull. ,2, p. 148, 1955. 



and their immediate or later issue. It is clear, then, that genetic 
uniformity of donors and hosts is not necessary for the success- 
ful use of homografts; what is necessary is that the donors 
should not possess any substances making for incompatibility 
which are not also present in the hosts. Mice of the first 
generation of a cross between the members of two highly inbred 
strains contain representatives of all the hereditary factors 
possessed by either parental strain; skin grafted from their 
parents or their progeny cannot therefore come to them, 
genetically speaking, as a surprise. All this is of great import- 
ance experimentally, but it has no bearing at all on practical 
everyday affairs, for human beings are, genetically, a most 
diverse assembly and even the most strenuous effort of abstrac- 
tion cannot liken them to inbred mice. 

The second exception to the rule that skin grafts are sure to 
perish after transplantation from one individual to another is 
this: it does not apply to embryos. Embryos will accept not 
merely homografts but grafts from members of quite different 
species as well, "heterografts". The age at which an animal 
becomes competent to recognize foreign tissue as foreign varies 
from one species to another. In sheep certainly,^ and almost 
certainly in cattle, the power to react upon and reject homo- 
grafts has already developed two-thirds of the way through 
pregnancy; in mice and chickens, the transition from the 
immature or embryonic to the adult mode of response is 
marked, to a good enough approximation, by birth itself. 
These variations are perfectly understandable: if we compare 
one species with another, it soon becomes clear that birth is a 
movable feast in the calendar of development, for mice are 
born at a stage of development not much different from that of 
a human foetus only three months old, and sheep and cattle 
are much more mature at birth than man. 

1 P. G. Schinkel and K. A. Ferguson, Australian J. Biol. Sci., 6, p. 533, 



The third exception, the surprising one, turns out on analysis 
to be a variant of the second. In some animals it is sometimes 
possible to exchange skin homografts between non- identical 
twins, that is between ordinary litter mates, which resemble 
each other in hereditary make-up no more closely than ordinary 
sibs. This dispensation applies to about 90 per cent of twin 
cattle — I must be understood to mean non-identical twins; that 
tissue homografts can always be exchanged between identical 
tAvins has already been conceded — and, so far as our present 
meagre evidence goes, to all twin chickens; and it can be 
assumed to be true of a certain very small proportion of twin 
sheep and a still smaller proportion, surely less than 0*1 per 
cent, of human twins. (It was for this reason that I put Victor ""s 
and Eric''s chances of being identical twins no higher than 
99*9 per cent.) Twin chickens, I should explain, are those that 
hatch from a two-yolked egg, each yolk being a separate egg 
as the embryologist understands that word. They could there- 
fore be described with equal propriety as uniovular or bin- 
ovular: it just depends on what one means by 'egg\ 

The non-identical twins between which it is possible to 
exchange skin homografts are among the most remarkable 
animals in nature, for they are graft-hybrids or chimeras; each 
twin is a mixture of cells of two genetic origins, most of its cells 
being its own, the remainder having been at one time the 
property of its partner. The exchange of homografts between 
them in later life does not therefore make them chimeras; it 
merely makes them more so. Chimerism of natural origin was 
first described by the American biologist R. D. Owen in 1945; 
nearly all cattle twins, so he found, contain a mixture, not 
necessarily a fifty-fifty mixture, of each other's red blood 
corpuscles. (This is known to be true of non-identical twins in 
cattle; it can be assumed to be true of identical twins, but it 
cannot be proved because their red blood corpuscles cannot be 
told apart.) How does this come about? The origin of the 



condition was clear enough, for most litter-mates in cattle are 
of the kind that share a common blood circulation from a very 
early stage of embryonic life until birth. All blood cells, there- 
fore, and all cells which may circulate in the blood stream on 
their way elsewhere, can be exchanged between the twins 
before they are born. But what is specially interesting is that 
the state of chimerism, of red-cell intermixture, may last for 
years or perhaps for Life — certainly for much longer than the 
lifetime of a red blood corpuscle, which is not likely to exceed 
a hundred days. It follows, then, that not merely red cells but 
the cells which make red cells must have been exchanged in the 
foetal cross-transfusion; and that, in defiance of the principles 
formulated in this article, they survived when the animals grew 
up and continued to manufacture the red corpuscles charac- 
teristic of their original owners. 

It is for exactly the same reason that twan chickens are red- 
cell chimeras, for the blood systems of the two separate 
embryos within the single shell communicate directly with each 
other. Moreover, my colleagues and I have shown, that the 
state of chimerism can be brought about artificially by joining 
two eggs together across a vascular bridge, using the method 
first devised by the Czechoslovak scientist Milan Hasek. 
Natural chimerism has been described in a small minority of 
twin sheep, and in one human being, a Mrs McK, who was not 
known to be twin when, at the age of twenty-five, she was found 
to contain a mixture of two genetically different types of red 
blood corpuscle.^ Her oa\ti Avere of blood O, but they were 
mixed with corpuscles of group A, the cellular relicts of a 
male twin who had died when three months old. There is 
no knowing how long Mrs McK will remain a chimera, 
but she has now been so for twenty-eight years; probably, in 
the long run, her twin brother'^s red blood cells will slowly dis- 

1 I. Dunsford, C. C. Bowley, A. M. Hutchison, J. S. Thompson, R. 
Sanger and R. R. Race, British Medical Journal j 2, p. 81, 1953. 



appear, and so pay back the still outstanding balance of his 

All known chimeras, then, are twins, and all such twins have 
been found to accept skin homografts from each other for as 
long as the state of chimerism endures. So when I say that 
embryos will accept homografts because they are not yet old 
enough to have acquired an immunological conscience, to have 
learnt the difference between what is native to them and what 
is foreign, that is only half the story. The other half is that 
foreign cells introduced into an embryo affect it in such a way 
that it may never acquire the power to recognize the cells as 
foreign, and may accept them as its own. It will accept, more- 
over, not merely the cells which gained access to it as an 
embryo, but any cell of the same genetic composition that may 
be transplanted to it in later life. This is the origin of the 
concept of acquired immunological tolerance, of which I shall 
say more in Section 5 below. 


Skin homografts are destroyed by an immunological re- 
action, that is by a process fundamentally akin to that which is 
provoked by bacterial, viral or cellular infections, or by the 
injection of foreign proteins or polysaccharides. For all the 
clinical good-will and perhaps even mortal urgency that accom- 
panies their transplantation, skin homografts are treated as if 
they were a disease of which their destruction is the cure. 
(This outdoes Erewhon: the disease is beneficial, the cure does 
harm.) Transplantation immunity therefore resembles the 
allergies and hypersensitivities and serum sickness, and trans- 
fusion accidents and haemolytic disease of the newborn, in 
being an immunological reaction-gone-wrong; and so far have 
we travelled from the days when all immunological reactions 
were supposed to be necessarily beneficial (however mysterious 



their benefactions might appear to be) that immunology is now 
hardly less commodious than psychology in providing an 
etiological funkhole for diseases of which the physical causes 
are still unknown. That most immunological reactions do good 
is of course a truism, though it is a truism Avhose truth has been 
formally proved only in very recent years. There exists a con- 
genital affliction, agammaglobulinaemia, the victims of which 
are unable or almost unable to manufacture blood protein of 
the class, gamma globulins, to which most antibodies belong. 
Being virtually unable to manufacture antibodies, sufferers 
from agammaglobulinaemia go down with almost every in- 
fectious disease they may be exposed to, and perhaps with the 
same disease again and again. Only antibiotics can keep them 
alive, to be witnesses to the truth that, under normal circum- 
stances, immunological reactions are necessary not merely for 
remaining in health but for remaining alive at all. The recog- 
nition of the disease in its severest form had therefore to await 
the discovery of antibiotics. The victims of agammaglobulin- 
aemia can be successfully grafted with foreign skin,^ and this 
has raised hopes that they might be surgically endowed with 
cells capable of manufacturing antibodies. 

But, it may be objected, how can we be certain that trans- 
plantation immunity is not a blessing in disguise — that it is 
not deeply harmful to mix up tissues of different genetic origins 
in a single individual, the very thing which transplantation im- 
munity normally makes it impossible to do? A few years ago this 
argument or innuendo might have carried weight, but now it is 
no longer tenable. Chimeras occur naturally or, as I shall explain 
later, can be made artificially. Compared with ordinary animals, 
chimeras are at no disadvantage; or, if they are at a disad- 
vantage (I am thinking of ''freemartinism'', the sterility of most 
female members of twin pairs of unlike sex in cattle), it is for 
reasons unconnected with chimerism as such. There is therefore 

1 11. A. Good and R. L. Varco, /. American Med. Assoc. ^ 157, p. 713, 1955. 



no danger that the attempt to make homografts permanently 
acceptable is going to brush aside some prudent natural safe- 
guard against compounding individuals of cells of different 
origins. There is only one special circumstance in which we 
need expect trouble: when the host will accept the graft but 
the graft will not accept the host. My colleagues have shown 
that this danger is not merely theoretical. 

Skin homografts are destroyed by what is technically known 
as an ""actively acquired** immunity reaction; there is no ready- 
made resistance against homografts, in the sense in which an 
individual of blood group O is ready equipped with antibodies 
capable of agglutinating red blood corpuscles from donors of 
groups A or B. Resistance to homografts develops in the course 
of, and as a consequence of, exposure to the foreign substances, 
antigens^ contained within them. At first, as I have already 
said, a skin homograft behaves just like skin merely transposed 
from one part to another of a single individual: it heals on just 
as soundly, it is as quickly and as richly re-equipped with a 
working vasculature, and it undergoes just the same processes 
of internal reorganization and repair. It may even survive long 
enough to grow new skin glands and a new crop of hair. But 
sooner or later, a reaction overtakes it. Just how soon that 
happens depends upon many variables, e.g. the quantity of 
foreign tissue that is grafted, for the more that is grafted, the 
sooner will it be destroyed. There is, however, one variable 
whose influence overrides all others, the genetical relationship 
between the donor and the host, and this is worth a momenfs 

Philosophers make a distinction between differences of 
degree and of kind, but the inborn differences between indi- 
viduals cannot be classified in either way. The differences 
between individuals are combinational, or, as mathematicians 
say, combinatorial differences; one individual differs from all 
others not because he has unique endowments but because he 



has a unique combination of endowments. The number of 
hereditary factors from which these combinations can be built 
up, though large, is finite, but the combinations themselves 
are far more numerous than the individuals who can enjoy 
them, so that for each man actually on stage there are hundreds 
of possible men still waiting for a cue behind the scenes. 

It follows that although the mechanism of heredity may be 
ultimately atomic — though, for reasons I shall explain below, 
I should prefer to describe it as a matter of chemical si7igu- 
larities rather than of physical particles — yet the relationship 
between human beings is defined by a virtually continuous 
spectrum of affinities, bounded at one end by identical twins 
and extending the other end far beyond the genetically visible 
region into the affinities between animals which are not 
members of the same species or even of the same order or class. 

The technique of skin grafting is particularly well qualified 
to demonstrate these propositions, for it can reveal {a) that all 
individuals, with the exceptions already noticed, are immuno- 
logically unique; (h) that the immunological differences between 
individuals are combinational in character; (c) that the com- 
binations are so diverse that there is an almost continuous 
range of variation in the acceptability of foreign tissues to their 
hosts. It is bounded at one end by grafts exchanged between 
identical twins, grafts which survive as long as the animals 
which bear them; and close to this end lie, for example, the 
grafts which have been transplanted from males to females of 
the same inbred line and which may survive their transplanta- 
tion by as much as fifty days. Grafts transplanted between 
ordinary members of the same species normally survive for 
little longer than a week; and when donors and hosts are 
members of different species, the grafts ('"heterografts'') never 
heal in properly, and it is only in a narrowly technical sense 
that they can be said to survive for any length of time at all. 
The range therefore extends from something near zero to 



something as near to permanence as mortality allows, and this 
variation is an expression, perhaps the completest single 
expression, of the genetical relationship between the donor 
and the host. 

A logical, or more properly a chronological, analysis of the 
reaction that leads to the rejection of homografts should begin 
with the antigens, the substances that cause immunity; proceed 
from the antigens to a consideration of whereabouts in the host, 
and how the reaction against them begins to take effect; and 
end \^ath an analysis of the effector mechanisms of immunity, 
i.e. of how the reaction of the host is, as civil servants say, 
actually *'implemented\ To avoid too many digressions, I pro- 
pose, however, to begin with the hosfs response and to leave 
any consideration of the antigens until the end. 

There is one thing that must be said immediately about the 
response of the recipient, for otherwise one or two of the 
experiments I shall describe below will be unintelligible. It is 
only when a human being or other animal is confronted with a 
homograft/br the first time that the homograft enjoys a latent 
period during which it behaves like a graft of the recipient''s 
own skin. A second graft from the same donor, transplanted 
after the rejection of a first, is set upon almost immediately; it 
does not heal properly, it never acquires a working vasculature, 
and it never even begins to reorganize itself internally or to 
develop new skin glands or hair. From a surgical point of view 
its destruction is virtually instantaneous, though its epithelial 
cells can survive a few days until they die of inanition. If, 
however, the second graft comes from a donor genetically 
different from the first, its behaviour may be almost completely 
normal. That is what one would expect, unless the donors of 
the first and second grafts happen to be closely related, in 
which case the second graft is summarily destroyed. This 
behaviour strengthens the analogy between an animaPs re- 
action against homografts and its reaction against a disease. 



When exposed for the first time to a homograft or a disease- 
producing organism, an animal takes the homograft or may get 
the disease. Resistance develops in the course of exposure — the 
disease is got over, the homograft sloughs away — and now, for 
many months, the animal becomes refractory, and will not get 
the same disease or take a graft of the same kind again. As to 
the specificity of the reaction, here too there is an exact 
analogy, for recovery from one disease will not prevent oner's 
succumbing to another unless, as with cowpox and smallpox, 
the organisms that cause them are closely related; and so it is 
with homografts, as I have just explained. 

It has been known for many years that bacteria which gain 
a foothold in the skin may enter the lymphatics, the system of 
vessels responsible for the fluid drainage of the tissues, and so 
enter the lymph nodes which lie athwart every lymphatic vessel 
somewhere between its source in the tissues and its final out- 
flow into the veins. Lymph nodes, less properly lymph ''glands'', 
are the organs in the neck and armpit and elsewhere which 
people refer to when they say their 'glands'" are swollen; and 
lymph nodes are probably the first places in the body in which 
antibodies are made. Antibodies made in one animal can be 
injected into another, so conferring upon it a vicarious, 
''passive'' or second-hand immunity. In medical practice, for 
example, antibodies against the toxins of tetanus or diphtheria 
organisms are commonly made in horses. The horse''s serum, 
now an antiserum, can then be injected into human beings. 

The reaction against homografts is much the same. Anti- 
genic substances from the homografts reach the local lymph 
nodes, normally via the lymphatics, and the local lymph nodes 
are the principal seat of the recipient ''s reaction. Mitchison^ 
has shown that if cells are taken from the lymph nodes of a 
mouse which has been actively immunized against a homograft, 
and if these cells are then injected into a second mouse of the 
1 N. A. Mitchison, Proc. Roy. Soc. B, 142, p. 72, 1954. 



same inbred strain, the second mouse behaves exactly as if it 
had itself been actively immunized beforehand, i.e. as if it had 
received and rejected a homograft before. This is analogous to 
passive immunization with an antiserum in the sense that it is 
the first mouse which undertakes the reaction against the 
homograft and the second which shares the benefit of it. But 
it is not exactly analogous, because the transference of the state 
of immunity cannot be brought about by injecting the first 
mouse''s blood or blood serum into the second; it must be the 
first mouse''s living lymph node cells. 

This is the first sign of an important difference between the 
immunity caused by bacteria (or other remotely foreign anti- 
genic substances) and the immunity caused by living cells 
originating from some other member of their recipient''s species. 
There can be no doubt that antibodies are the chief instru- 
ments of the defensive reaction against what may be com- 
pendiously described as ""germs'. It is true that antibodies 
themselves do not seem to do bacteria much harm; what they 
do is to make the bacteria particularly palatable to phagocytes, 
or to make them sensitive to the action of a complex con- 
stituent of the blood known as 'complemenf, which dissolves 
them. But although antibodies may not be sufficient to bring 
about the destruction of bacteria, they are certainly necessary; 
yet in the reaction against skin homografts there is no clear 
evidence that they are even necessary. 

The highly skilled researches of Dr P. A. Gorer of Guy''s 
Hospital have shown that antibodies are certainly formed when 
a homograft of skin or of tumour cells is reacted upon and 
sloughed away. The antibodies are of a chemically quite 
orthodox kind, and may be recognized by their power to 
agglutinate the red blood corpuscles of the donor. If antibodies 
are formed, why should we doubt that they are the chief 
eff'ectors of the immunological response? 

The main reason, perhaps, is that a state of immunity cannot 




be transferred from one mouse to another by transfusions of 
blood or serum, for all that the blood may contain a high 
concentration of antibodies. A second is that the blood of an 
animal which has received and rejected a homograft contains 
nothing that opposes or even discourages the growth of a 
donor''s cells in tissue culture. A third has emerged from some 
ingenious work carried out at the National Cancer Institute 
in the U.S.A.^ If a donor^s cells are grafted into an animal 
which has been forearmed against them, by having received 
and rejected cells of the same origin before, they are promptly 
set upon and destroyed. But if the donor's cells are housed 
inside a little permeable plastic bag within the recipient, then 
they will be destroyed if, and only if, the walls of the bag are 
permeable enough to let through the recipient's cells. Perme- 
ability to molecules of the size of antibody molecules is not 
enough. Clearly, then, it is not sufficient merely to confront a 
donor''s cells with antibodies. Is it even necessary to do so.'^ 
Apparently not, for the following reason. A donor's tissue, 
lying in a plastic bag that is permeable to antibodies but not to 
cells, will be destroyed if cells from the immunized recipient 
are added to the contents of the bag before it is sealed up and 
introduced into the recipient. This experiment reproduces the 
state of affairs in which the donor's tissue is housed in a plastic 
bag that is permeable to the recipient's cells, the only difference 
being that, in this variant, the recipient's cells are spared the 
exertion of getting in. But the donor's tissue is also destroyed 
if the bag and its contents are grafted into the donor or an 
animal genetically similar to the donor. The donor's body 
fluids may well contain something necessary for the immuno- 
logical reaction to take effect within the plastic bag, but what- 
ever that may be, it cannot be an antibody, for the donor 
cannot contain antibodies acting against its own cells. 

1 J. M. Weaver, G. H. Algire and R. T. Prehn, /. National Cancer Inst., 
15, p. 1737, 1955. 



For these and other reasons, most of us are now convinced 
that antibodies play no necessary part in the reaction that 
destroys skin homografts and most other homografts of soUd 
tissues. With homografts of isolated cells, and more particu- 
larly of leukaemic tumours, it may be a different story. By 
'antibodies'* I must be assumed to mean the ordinary, orthodox 
antibodies that circulate freely in solution in the blood stream 
and can be recovered in serum free from cells. Perhaps the cells 
that do destroy homografts are transporters of antibodies and 
perhaps they liberate them exactly where they can do most 
damage, that is, in the immediate neighbourhood of the grafts; 
but if that proves to be the case, I suspect that the antibodies 
will turn out to be so far different from ordinary antibodies as 
to deserve some different name. 

I now come back to the problem Avhich I should have begun 
with: what are the antigens, i.e. the substances in homografts 
that actually set the immunity reaction going? This may be 
said at once: that whatever the antigens are, there are a great 
many of them and they are under very exact genetical control. 
So much was made clear by the pioneer work of C. C. Little, 
developed to a remarkable degree of refinement by the school 
of research which he created in Bar Harbor, Maine (notably 
by G. D. Snell), and by P. A. Gorer at Guy''s Hospital. But the 
researches which have made it possible for antigens to be 
labelled, traced through their Mendelian evolutions, and 
separated by purely genetical methods has left entirely open 
the question of what they are. Without recourse to any kind 
of physical or chemical analysis, it can at least be said that the 
antigens cannot be substances which are in any way peculiar 
to any one kind of living cell. All the antigens belonging to 
that combination which distinguishes a particular individual 
are to be found in all the living tissues of which his body is 
composed. Analysis by methods making use of the principle of 
'tolerance** (Section 5) shows that the skin cells of a particular 



mouse contain no "transplantation antigens'* not also present 
in its kidney cells or leucocytes or spleen. The hall-mark that 
distinguishes a particular individual is to be recognized in all 
his living cells and must therefore be stamped upon some 
constituent which all cells share in common. What, then, is the 
chemical nature of the substances which leave a homograft, 
enter the lymphatics, reach the regional lymph nodes and there 
set in train the reaction that ultimately causes the homograft 
to be destroyed? 

On the face of it, such a problem should not be too difficult 
to solve; yet it is not unjust to say that fifty years of research 
upon the problem of incompatibility had not, until very 
recently, provided us with even the beginnings of an answer. 
For the antigenic substances are grievously unstable. If a 
living tissue is heated for a few minutes to a temperature of 
49° C. (which is not so much hotter than a hot bath), or thrice 
alternately frozen solid and then thawed, or frozen solid and 
then dried under a high vacuum in the frozen state, it loses its 
power to provoke immunity. All these treatments kill cells, 
but they do so in the humanest possible way, i.e. in the way 
that does the least possible unnecessary damage to their fine 
structure — and cells must surely be killed if they are to be 
separated into their several chemical or anatomical fractions. 
So we had to resign ourselves to a belief belonging conceptually 
to the dark ages, viz. that the power to cause transplantation 
immunity was a prerogative of living cells. 

The findings w^hich I am now about to describe^ began with 
the discovery that it is possible to kill and disintegrate cells 
without destroying the substances that cause transplantation 
immunity. It is done by exposing cells to very loud and very 
high-pitched sound; in effect, then, to vibrations of high 
ampUtude and frequency. The amplitude of the vibrations 

1 R. E. Billingham, L. Brent and P. B. Medawar, Nature, 178, p. 514, 

L 161 


which will shatter cells to pieces in less than one minute is that 
corresponding to a power output of about 50 watts, which, I 
suppose, is about the power output of a large symphony 
orchestra when giving its patrons full value for money. The 
frequency is not a matter of great moment; something between 
five and ten octaves above Middle C will do, i.e. a frequency 
extending from just above that which is audible to human 
adults to that which would correspond to a piano key lying 
three and a half feet beyond the top end of the piano. Cells 
which have been exposed to ultrasonic irradiation are shattered 
into minute fibres or rounded particles, but on injection they 
can still cause transplantation immunity, i.e. they can cause an 
animal to behave towards a homograft exactly as if it had 
received and rejected a living homograft before. 

The next problem was to find out whereabouts in the cell 
the antigenic matter was located — whether in the nucleus (an 
extremely far fetched possibility) or in the remaining substance 
of the cell. We showed beyond doubt that the antigens were 
confined exclusively to the nucleus. More than that, we have 
good evidence (not at this moment quite conclusive) that the 
antigenic substances are desoxyribonucleoproteins. The im- 
portance of this discovery is that desoxyribonucleoprotein is 
the chemical nexus of heredity, and is the stuff of which 
chromosomes are made. 

A nucleoprotein is a salt-like compound between nucleic 
acid and protein, mainly basic protein. If it is to fulfil its 
genetical function, one or other of its two ingredients, if not 
both, must be sufficiently commodious and versatile to act as 
the scrupulously exact and detailed invoice of all the inborn 
diversity of mice or of men. Modern opinion inclines to the 
view that it is the nucleic acid which fulfils this immeasurably 
important function. The nucleic acids have been recognized as 
chemical entities for more than sixty years, but the crucial 
observation which began to shape the modern opinion of their 




importance was not made until 1928. In that year Fred. 
Griffith, a bacteriologist employed by the Ministry of Health, 
wrote a carefully reasoned paper in which he recorded the 
following surprising fact: that if a mouse were inoculated with 
a mixture of living pneumococci of one type and dead pneumo- 
cocci of a second and quite different type, there would arise in 
the mouse living pneumococci of the second type. Looking 
back, we can now see that he had accomplished something far 
more important and richer in possibilities than the transmuta- 
tion of elements; in principle at least, he had accomplished the 
transmutation of a biological species. Others pursued his 
discovery. First, it was found that the transformation could 
happen in a test-tube, not necessarily in a mouse; then, that 
pneumococci of one type could be transformed into pneumo- 
cocci of a second type by an extract from the latter, i.e. not 
necessarily by whole dead cells. O. T. Avery and his colleagues 
have since sho^\Ti that the transforming principle is a desoxy- 
ribonucleic acid, a nucleic acid of the type which, in higher 
organisms, is confined exclusively to the nuclei of cells. Up- 
wards of twenty such examples of genetic transformation are 
now known, each one mediated by a particular nucleic acid. 
The nucleic acid behaves as if it were a ''gene'': it enters the cell 
by a process analogous to infection, takes its place among the 
other determinants of heredity, and like the other determinants 
is continually reproduced anew. Viruses and bacteriophages 
are also nucleoproteins and they too act essentially by bringing 
about a cellular transformation. It is probable that the trans- 
formations brought about by viruses are essentially similar to 
those produced by a solution of nucleic acids; indeed, when a 
bacteriophage enters a bacterial cell, it is believed to leave its 
protein moiety outside, as if the protein were the mortal part 
of its constitution; only the nucleic acid goes inside. 

For these weighty reasons, it is now widely believed that the 
nucleic acid moiety of the nucleoprotein molecule is that which, 



in modern jargon, acts as the carrier of genetic information. 
But what, or of what nature, is the code? Nucleic acids are 
essentially polymers, or repetitive manifolds, of simpler units 
known as nucleotides, and each nucleotide is a compound 
between phosphoric acid, a sugar molecule with five carbon 
atoms (unlike glucose, which has six, and ordinary cane sugar, 
twelve), and an organic base which belongs to one or other of 
two families, the purines and pyrimidines. In desoxyribonucleic 
acid, the one we are concerned with, there appear to be four 
possible bases, if we count two closely related bases as one: the 
two purines are adenine and guanine, the two p}Timidines 
thymine and cytosine. The constituent nucleotides are so 
arranged that the sugar group of one attaches to the phosphate 
group of its neighbour; the backbone of the polymer is there- 
fore an alternating sequence of phosphate and sugar groups. 
But what is all important is their arrangement. According to 
the analyses of Watson and Crick,^ helped by the X-ray 
diffraction studies carried out at King''s College, the molecule 
of desoxyribonucleic acid is a pair of threads, each having the 
polymeric structure just described. The threads lie equidistant 
from each other, and are twisted to form a double spiral; and 
(still following Watson and Crick) matters are so disposed that 
the adenine base of the one thread is linked to a thymine base 
on the other, and the guanine of one to the cytosine of the 
other. The two threads are therefore of complementary struc- 
ture: given the sequence of bases on one, the sequence on the 
other is automatically fixed. This arrangement makes it 
possible to envisage how the molecule is reproduced, for if each 
of the two constituent threads builds up a complementary 
thread upon itself, the four may now split lengthwise along the 
line of union between the original two, so giving two double 
threads, identical with each other and with the double thread 
from which they first arose. But this does not solve the problem 
1 J. D. Watson and F. H. C. Crick, Nature, 171, p. 964, 1953. 


■ r 


of how SO simple a molecule can embody a genetical programme, 
the huge variety of detailed and exact instructions or proposals 
which are carried out, or disposed, as development unfolds. 
If this interpretation of the structure of nucleic acids is correct, 
they can differ one from another only in the ordering of the 
bases along the length of the chain — a profoundly revolution 
ary view, because it puts an end to the atomic or particulate 
conception of the '"gene\ A gene can still be thought of as a 
particle, but it can no longer be a particle; it is now an ordinal 
singularity of a nucleic acid molecule, a region in which some 
unique and distinctive permutation of the organic bases prevails. 

The antigens that cause transplantation immunity, I have 
said, are probably desoxyribonucleoproteins. It is as if the 
antigens were genes. I shall now examine three consequences of 
this interpretation, the first two immediate and concrete, the 
third intangible and vague. 

The first is this. In theory, one way of making a homograft 
acceptable to a host which would otherwise have rejected it 
would be to alter its antigenic constitution, so that it could no 
longer provoke an immunological reaction from the host. If the 
antigens are nuclear nucleoproteins, this solution is out of the 
question, for the antigens are built into the inner genetic 
substance of the cell. 

The second consequence is rather more encouraging. We are 
beginning to learn something about the structure of nucleic 
acids, but of the exact nature of the union between a nucleic 
acid and a protein that makes a nucleoprotein we are still in 
doubt. If the antigens of transplantation immunity are nucleo- 
proteins, tissue grafting should provide us with a direct test of 
their biological integrity, that is, of the degree to which, after 
extraction and chemical handling, they retain the properties of 
their native state. The trouble with the nucleoproteins ex- 
tracted from higher organisms is that nobody has so far been 
able to do anything with them except record and ponder 



uneasily upon the significance of their physical and chemical 
properties; but the only way to be sure that they are still in 
working order is to see if they can still discharge some genetic- 
ally specific function. The study of nucleic acids was in the 
same condition, all dressed up and nowhere to go, until the 
development of Griffith's great discovery of bacterial trans- 
formation. It is to be hoped that skin grafting will do as much 
for nucleoproteins, for in studying the antigenic activity of 
homografts we are studying a proximate genetic property, and 
it is very likely that anything which destroys the power of a 
nucleoprotein to act as an antigen in transplantation im- 
munity will also destroy its genetic competence, i.e. the pro- 
perties which empower it to direct the course of development 
along certain exactly specific lines. 

The third implication has not yet shaped itself into any- 
thing solid enough to be called an opinion; it is still in the 
gestation period of conscious thought, a state of querulous 
unease. In Section 5 of this article I shall give reasons for 
believing that antigenic substances of the kind that cause 
transplantation immunity are constantly manufactured and 
emitted during the everyday life of ordinary living cells. They 
give no evidence of their presence or activity until the tissues 
which manufacture them are grafted upon an individual to 
whom they do not belong. They then act as antigens, setting in 
train a reaction which causes their source, the grafted tissue, 
to be destroyed. What then is the normal function of these 
substances which act as antigens in the entirely artificial 
context provided by grafting? There is every reason to believe 
that all the nucleated cells of the body give forth these nuclear 
particles, each carrying the genetic hall-mark of the individual, 
and that they enter the lymphatics and in due course reach the 
regional nodes. What, if anything, do they do? 

It may be said at once that although the continued life of 
the cell may well depend on the continual emission of nucleai 



matter, it certainly does not depend upon any use which the 
matter may later be put to. If it did, the life of the cell outside 
the body, in tissue culture, would not be possible. Perhaps the 
nuclear substances discharged by cells have to do with the 
control of growth, for tissues maintained by culture and sub- 
culture outside the body, in media constantly renewed, indulge 
in unlimited growth; perhaps the reaction which occurs when, 
in homografting, nuclear substances of slightly the wrong kind 
enter the lymph nodes and other centres of response is a tell- 
tale aberration of some deeply important regulatory mechan- 
ism, essentially of an immunological character, by which these 
substances are dealt with in ordinary life. Several scientists 
have thought along such lines and are seeking evidence of an 
essentially immunological control of growth. Might not the 
nuclear particles I am discussing be its agents? 

Unfortunately, I do not think these nuclear particles will 
fill the bill. Substances which control the growth of particular 
tissues — of one tissue, that is, but not necessarily of another — 
must surely be specific to and characteristic of those tissues; 
but within any one individual the nuclear antigens are common 
to them all. I incline to the more homely view that the emission 
of nuclear particles is excretory in nature. No one doubts that 
the nucleus administers the activity of the outl3dng parts of 
the cell, but no one knows exactly how it makes its wishes 
known. It is likely (though this is only one possibility among 
several) that genetically specific matter leaves the nucleus and 
enters the cytoplasm, though not at all likely that, having done 
so, it goes back; it may simply be discharged through the 
surface of the cell, eventually to reach the lymph nodes or other 
clearance centres and to be broken down. Something of the 
kind has long been thought to happen in nerve cells, for the 
integrity and working order of a nerve fibre is absolutely 
dependent upon the presence of the nucleus in the cell-body 
from which the fibre arises. The dependence of the nerve fibre 



upon its nucleus is doubly instructive, (a) because in large 
animals the nucleus may be several feet away from the most 
distant part of the fibre, so that the idea of something leaving 
the nucleus, reaching the farthest tip and then returning home 
becomes almost impossible to credit: the traffic of nuclear 
particles can only be one-way; and (b) because it gives the 
clearest possible evidence of the activity of the nucleus during 
its so-called ''resting stage**, i.e. between one nuclear division 
and the next. The nuclei of mature nerve cells do not divide, 
so that their labours, though doubtless strenuous, are all 
sabbatical. Fortunately, there is other evidence that nuclei 
are busy in the intervals between dividing; for example, nuclei 
isolated from cells seem to be able to incorporate substances 
from the environment by active metabolism and to enjoy 
considerable synthetic powers.^ Yet their own substance does 
not increase until just before the act of division, so that what 
is made in the nucleus must presumably go elsewhere. I 
emphasize these points, perhaps to the extent of labouring 
them, only to make clear that the idea of nuclear substances 
entering the cytoplasm and then being discharged through the 
surface of the cell is not at all surprising. What is surprising 
is that a technique like skin grafting should provide us with 
evidence that it occurs. 


It is at present possible to envisage four kinds of ways in 
which a homograft could be made acceptable to its host: (a) the 
antigenic constitution of the graft might be changed, so that 
it no longer stirred up a reaction in its host; (b) the graft might 
be transplanted in such a way that it could not exercise its 
antigenic properties; (c) the host might be changed in such a 
way that its reaction against the graft was enfeebled or done 

1 V. G. Allfrey and A. E. Mirsky, Nature, 176, p. 1042, 1956. 



away with altogether; or (d) the graft might be put in a position 
in which, no matter what state of immunity might prevail, it 
could not be got at by the cells that put the immunity into 

The first solution cannot be applied to homografts of the 
kind I have been particularly concerned with, grafts which are 
alive when they are grafted and which must remain alive if they 
are to do their recipient any permanent good. Why this should 
be so has already been explained: the antigenic make-up of a 
graft is built into its genetic constitution. I therefore grieve at 
the theoretically infirm attempts which have been made to 
change the antigenic constitution of a graft by, for example, 
growing it as a tissue culture in the body fluids of the animal 
or human being on which it is ultimately to be transplanted — 
accustoming it gradually (such is the feeble hope) to what it 
will have to make do with later. No antigenic transformation 
is in the least likely to occur under these conditions. Antigenic 
transformations can occur under very special conditions which 
have no bearing on the way in which homografts are used in 
surgical practice; for example, if a graft consisting of isolated 
tumour cells is transplanted to an animal which puts up a 
certain feeble resistance to its growth, then the population of 
tumour cells, considered as a whole, may change its antigenic 
properties; but that I conceive to be due to a process of natural 
selection, i.e. the selection, from a rapidly growing and prob- 
ably variable population, of the particular variants that are 
least antigenic to the host. 

Not all grafts are of the kind that need remain alive after 
their transplantation; homografts of segments of blood-vessels, 
for example, are building up an impressive record of successful 
use in human surgery, but, to put it as a paradox, they are 
successful as homografts because their failure does not matter. 
What a vascular homograft does is to provide a fibrous tube 
of the right shape and texture which, when its own cells die, 



is repopulated by cells arising from its recipient; a new lining 
of endothelial cells is laid down on the inner surface of the 
graft, and cells of the connective tissue family penetrate the 
interstices of its fibrous skeleton and convert it into a plausible 
and efficient imitation of a normal blood vessel. That vascular 
homografts, as living entities, '"die'' and are none the worse for 
it is shown by the fact that they need not be alive even to 
begin with. Vascular homografts which have been killed 
(though kept in a lifelike condition) by drying from the frozen 
state or by prolonged storage at very low temperatures seem 
to do as well as living grafts; synthetic plastic tubes will also 
serve. Vascular homografts die, but they enjoy the privilege 
of reincarnation. 

The second and third ways of getting round the homograft 
reaction are complementary to each other; either to transplant 
a homograft into a position in which the antigens manufactured 
by it never reach a centre of response, so that the host is never 
officially aware of its existence, or to transplant it into a 
position in which the effectors of the immunological response 
are unable to get at it. The main seat of the hosfs reaction 
against homografts, I have explained, is the regional lymph 
nodes, and antigens reach them through the regional lymph- 
atics. The brain has no lymphatic drainage in the ordinary 
sense, nor is it monitored by lymph nodes. It is therefore 
entirely intelligible that homografts transplanted into the 
substance of the brain should fail (as they do indeed fail) to 
elicit an immunological reaction. 

The complementary case is best exemplified by the cornea. 
Homografts transplanted into the cornea are in a kind of 
sanctuary. Their condition is to be likened, perhaps, to that of 
homografts transplanted into the little plastic bags, permeable 
to fluids but not to cells, which I mentioned in an earlier sec- 
tion. The cornea is a non-vascular structure, so that the cells 
which are the effectors of the hosfs reaction against homo- 



grafts simply cannot get through it — unless, indeed, the cornea 
should become vascularized by accident, in which case a graft 
transplanted to the cornea will usually fail. Brain and cornea 
are therefore both privileged sites of grafting, but for entirely 
different reasons. A simple experiment will make the distinc- 
tion clear. A homograft in the brain will certainly be destroyed 
if the animal into which it is transplanted is immunized by 
some other, efficacious route — for example, by a skin homo- 
graft transplanted upon or beneath the skin. It is therefore 
entirely vulnerable to an immunological reaction; it owes its 
privilege only to the fact that it cannot set such a reaction 
going. But a homograft in the cornea, provided the cornea is 
unvascularized, will survive in the face of a fulminating 
immunity directed against it; it survives because the state of 
immunity cannot take effect. 

So much for what may be called the bye-laws of tissue 
transplantation, the special rules that govern the behaviour of 
homografts in special positions in the body. I have mentioned 
two or three, but others remain to be discovered. No one has 
yet put forward a plausible explanation of why it is that homo- 
grafts of certain endocrine glands — of the ovary, for example, 
in so far as it is a source of hormones, or the adrenal cortex — 
sometimes survive when homografts of skin demonstrably fail 
to do so. But a general solution of the homograft problem must 
turn upon the last of the four expedients which I mentioned in 
the introduction to this section, i.e. upon changing the host 
in such a way that its reaction against homografts is done away 
with or at least enfeebled. 

There are half a dozen ways in which the intended recipient 
of a homograft can be treated in order to prolong the homo- 
graft ""s normal lease of life; most are temporary, but one can be 
permanent; some could be applied in surgical practice if it were 
worthwhile doing so, others not; a few are innocuous, but most 
are harmful. By far the most important distinction, however, 



is between treatments which are non-specific and treatments 
which are specific in their action. By a non-specific treatment 
I mean a treatment which will weaken, or under extreme 
circumstances abolish, the reaction against homografts from 
all sources — and, for full measure, probably abolish most other 
immunological responses as well. A specific treatment is one 
which weakens the reaction against homografts from some one 
particular donor, or from the members of some one highly 
inbred strain, without prejudice to the reaction against homo- 
grafts from other sources or, a fortiori, to the recation against 
antigens of other quite different kinds. 

I shall mention only two of the non-specific treatments: 
X-irradiation, and the injection of cortisone; and I mention 
X-iriadiation not because it is useful in the surgery of trans- 
plantation but simply because it has the awful prestige of 
anything to do with the threat of atomic war. ''Whole body 
irradiation"* of a sufficient dosage — for a mouse, something less 
than one thousand Roentgen units — causes, amongst other 
things, complete immunological prostration and severe and 
usually irreparable damage to blood-forming cells. It was dis- 
covered in America that mice which had received a dosage of 
radiation which would otherwise have been rapidly fatal could 
be kept alive by injecting them with cellular pulps made from 
the blood-forming tissues of other mice or even of members of 
alien species. For many years the nature of the protective agent 
contained within this pulp remained in doubt; some main- 
tained that it was humoral in nature, and that it could exist 
apart from living cells, others that the injection of the pulp was 
in effect a transplantation of normal living cells which simply 
took the place of those damaged or destroyed by radiation. 
Two independent groups of scientists,^ one at Harwell and the 

^ C. E. Ford, J. L. Hamerton, D. W. H. Barnes and J. F. Loutit, Nature, 
177, p. 452, 1956; D. L. Lindsley, T. T. Odell and F. G. Tausche, Proc. Soc. 
exp. Biol. Med., 90, p. 512, 1955. 



other at Oak Ridge, have now shown that the latter explana- 
tion is certainly correct. A heavily irradiated mouse is no 
longer capable of resisting the transplantation of foreign, even 
of remotely foreign, tissue; when it is injected with blood- 
forming cells from bone marrow or other blood-forming centres 
of normal mice, the cells establish themselves without opposi- 
tion in their new surroundings, and the irradiated mouse 
becomes a '"radiation chimera** in which the foreign blood- 
forming tissues act proxy for its own. 

The injection of high doses of cortisone, a drug which is 
closely related to one of the natural secretions of the cortex of 
the adrenal gland, produces an effect which has something in 
common with radiation sickness. Steroid hormones of the class 
to which cortisone belongs have a powerfully inhibitory effect 
upon the growth and activity of all lymphoidal tissue — upon, 
therefore, the cells which undertake their owners' immuno- 
logical reactions. The injection of large doses of cortisone can 
certainly prolong the life of homografts, though (for reasons 
which are still not quite fully explained) it does so more readily 
in mice and rabbits than in guinea-pigs or human beings. At 
one time we hoped that cortisone would be a useful minor 
addition to the armoury of the plastic surgeon in the treatment 
of very extensive burns. The great raw wounds left by deep and 
widespread burns are still sometimes covered with homografts 
of skin; homografts make a perfect temporary dressing which 
may tide the patient over until he can afford to provide some 
skin grafts of his own. It would buy useful time if these homo- 
grafts were made to survive only twdce as long as could 
otherwise be expected, and this is what we hoped that cortisone 
would do. Unfortunately, it seems that cortisone could prolong 
the life of homografts on human beings only at dosages which 
would have secondary ill effects of a gravity which an already 
sick patient could not put up with Yet the research account 
is by no means all a matter of debit, for study of the action of 



cortisone and other steroid hormones on the behaviour of 
homografts, particularly in its recent developments by P. L. 
Krohn, is giving us a new insight into the nature of the normal 
adrenal cortical secretions, how they vary from members of one 
species to another, and how they are influenced by the trophic 
hormone of the pituitary gland. 

The weakening of an animaPs reaction against homografts 
which can be brought about by the treatments or maltreat- 
ments I have just described is simply a by-product of some 
more general biological damage. The specific treatments I now 
turn to are nicely discriminating in their action; they influence 
the survival of homografts from particular donors chosen 
beforehand, and have no efl'ect on other immunological re- 
actions. One such treatment, the particular study of research 
workers in the Roscoe B. Jackson laboratory at Bar Harbor, 
Maine, entails the injection of the animals which are later to 
receive homografts with desiccates or extracts made from the 
tissues of their future donors. The theoretical importance of 
this treatment outweighs its practical usefulness, v/hich by all 
appearances is very modest, for the best it has been able to do 
in our hands is to double or treble the normal expectation of 
life of a skin homograft. Beyond the fact that it is certainly 
immunological in character, there is no common agreement 
about the way this treatment works. My colleagues and I are 
inclined to interpret it as an interference with the normal 
reaction against homografts which is somewhat analogous to 
*'desensitization*' against an allergic state. Allergic symptoms 
of the kind caused by, for example, pollen, are thought to be 
due to the union of the pollen antigens with a peculiar and 
unstable kind of antibody known as a *'reagin\ According to 
one interpretation, which may be unduly simple, desensitiza- 
tion consists in inducing the body to form orthodox and 
inoff'ensive antibodies against pollen which, by getting in first, 
cover or coat or otherwise preoccupy the pollen antigens so 



that they can no longer combine with the antibody that does 
the harm. Mice which have been injected with extracts or 
desiccates of their future donor''s tissues certainly do produce 
orthodox serum antibodies in large amounts — antibodies 
which, according to the reasoning of an earlier section, are not 
the instruments of the homograft reaction; and we think it 
possible that these antibodies have the power to combine with 
or otherwise inactivate the antigens which cause transplanta- 
tion immunity, perhaps as they issue from the grafts. 

The second '"specific"* method of interfering with the homo- 
graft reaction is that which turns upon the principle of immun- 
ological tolerance, and it deserves — or at all events is to receive 
— a chapter to itself. 


When antigens are injected into juvenile or adult animals, 
they provoke some kind of immunological response. That is not 
an empirical fact but a tautology: "antigens"* are so defined. It 
is an empirical fact, however, that when antigens are injected 
into embryos, or into newborns of the kind that are born very 
immature, they do not elicit an immunological reaction. For 
many years immunologists were content to dismiss this fact by 
saying that the immunological faculty is one that develops and 
matures like any other, and that embryos do not react upon 
antigens simply because they are not yet sufficiently grown up. 
This is a half truth; the other half of the truth is the subject 
of the present chapter. 

In 1949, F. M. Burnet and Frank Fenner propounded a 
theory of antibody formation which led them to make the 
following prediction: that if an embryo were to be injected 
with an antigenic substance, then, when it grew up, its power to 
react against that antigen would be found to have been 
seriously impaired. Almost all they had in the way of hard facts 



to go on was Owen's discovery (p. 150) of red-cell chimerism in 
twin cattle — the state of affairs in which cattle twins are born 
with and long retain a mixture of each other''s red blood 
corpuscles, presumably because they exchange blood-forming 
cells in embryonic life. If the exchange had not occurred before 
birth, but had been carried out artificiallv afterwards, then 
the foreign blood-forming cells would quite certainly have been 
recognized as such and destroyed by an immunological reaction. 
The exchange of the cells before birth must somehow have pre- 
vented the development of that faculty which would have em- 
powered the twins to recognize each other*'s cells as not their own. 
My colleagues and I have shown that Burnet and Fenner''s 
prediction is true, without qualification, of the antigens which 
are responsible for transplantation immunity. We too began 
our work on cattle. In 1948, while attending an International 
Congress of Genetics at Stockholm, I was invited by Dr H. P. 
Donald to help to solve the important problem of distinguishing 
with complete certainty between identical and non-identical 
twins in cattle. In principle, nothing could be easier. Skin 
grafts were to be exchanged between the twins a few weeks 
after their birth. If the homografts survived, the twins could 
be classified as identical; if not, as non-identical, i.e. dizygotic. 
My colleague Dr Billingham and I, helped by two young officers 
of the Agricultural Research Council, began what was to be a 
few months'* work later on that year; as it turned out, the work 
took three years to finish. We satisfied ourselves that the 
reaction against skin homografts was no less vigorous in cattle 
than in man or in laboratory animals, and that skin grafts 
exchanged between cattle of the same or of different breeds, 
or between dam and calf or vice versa, or between ordinary 
siblings (brothers or sisters but not twins) were all destroyed 
within a fortnight of their transplantation. But homografts 
between almost all the twins survived, irrespective of whether 
the twins were identical or dizygotic. There could be no mistake 



about the classification of the twins as dizygotic, for the pairs 
we chose for our critical tests were of unlike sex and as different 
as possible in other ways. Dizygotic twin calves were therefore 
tolerant of homografts of each other's skin — though not, it 
should be observed, of the skin of any other cattle. Obviously 
the cross- transfusion in foetal life which led to their becoming 
red-cell chimeras had destroyed their power to recognize each 
other*'s skin as foreign. Simonsen later reported the successful 
exchange of whole kidneys between dizygotic cattle, so it is 
likely, as more recent experimental work has shown to be true 
of mice and other laboratory animals, that dizygotic twin cattle 
are tolerant of homografts of aZZ tissues from their twins, though 
of no tissue from any other animal. 

Billingham, Brent and I therefore set ourselves to reproduce 
experimentally, in laboratory animals, the state of affairs that 
occurs by a felicitous natural accident in twin cattle and in 
those other natural chimeras mentioned in Section 2; and after 
a year's labour we succeeded. ^ A typical experiment was con- 
ducted thus. The seventeen-day old embryos of white mice of 
strain A were injected, while still i7i utero, with a mixture of 
cells taken from an adult donor belonging to the quite different, 
brown, strain of mice, strain CBA. The injected mice were born 
a few days after their injection, and allowed to grow up. A 
normal adult mouse of strain A rejects skin homografts from 
CBA mice within eleven days of their transplantation, but adult 
mice which had been injected before birth with CBA cells were, 
in the extreme case, completely tolerant of grafts transplanted 
from CBA donors. Graft hybrids could therefore be made at 
will. Since these first experiments were done, the technique has 
been greatlv simplified — with some strains of mice, for example, 
the preparatory injection can be delayed until immediately 

1 A full account of this work is contained in R. E. Billingham, L. Brent 
and P. B. Medawar's monograph in Philos. Trans, Roy. Soc. B, 239, 
p. 857, 1956. 

M 177 


after birth — and has been extended to other mammals and 
to birds. 

It follows, then, that the *"homograft reaction'' can be com- 
pletly abrogated; the problem of making animals completely 
tolerant of foreign tissue is not, as I had once feared, insoluble. 
But it must be said at once that it is hardly feasible to apply 
this technique to human beings — not so much because it would 
require interference with an unborn baby, though that objec- 
tion is grave enough, as because the state of tolerance is 
absolutely specific. The white mice I referred to above, made 
tolerant of CBA grafts by injecting them before birth with CBA 
cells, invariably reject skin homografts from other, unrelated 
donors; so likewise a human foetus injected with (for example) 
blood cells from Mr Smith (il s''agit ici de pseudonymes) could 
not be expected to accept homografts in later life from anyone 
except Mr Smith himself. 

Among the multitude of experiments which my colleagues 
and I have done in course of our analysis of the phenomenon of 
tolerance, two only will be singled out for special mention. The 
first is that the state of tolerance does not discriminate between 
the different tissues of a single individual; an injection of blood 
into the embryo will cause a tolerance of skin, an injection of 
spleen cells will cause tolerance of a graft of the cortex of the 
adrenal gland, and so on. If a tissue A, injected into an embryo, 
is to cause tolerance of some other tissue B, grafted later in 
life, then clearly B must contain no antigen that is not also 
present in A. (If B had some antigen peculiar to itself, there 
is no reason why it should not go into action and immunize 
the host.) This condition appears to be fully satisfied when A 
and B are any two different tissues from the same individual. 
It follows that all the living tissues of a single individual must 
have the same antigenic composition, w^hich is exactly what we 
should expect if the antigens do indeed consist of chromosomal 
matter. This is of great practical importance. In attempting 



to devise methods of overcoming the homograft reaction by 
techniques of desensitization, many unwary workers have been 
guided by the quite unfounded behef that the several tissues of 
the body contain ''homograft antigens'* pecuhar to themselves. 
To make a skin homograft survive, therefore, it was thought 
necessary to inject its intended recipient with some desensit- 
izing preparation made from skin itself. The basis of this judge- 
ment can now be seen to be quite illusory. If ever it becomes 
possible to desensitize an adult animal for the purpose of 
executing a homograft, it will doubtless be done by injecting 
it beforehand with some preparation of its future donor''s 
tissues; but the tissue used for that purpose need not be of the 
same kind as that which is ultimately to be grafted; blood 
leucocytes should serve the purpose as well as any other kind 
of living cell — a most important dispensation, because blood 
is of all tissues the easiest to come by and the easiest to spare. 
The second property of tolerance I wish to mention also 
bears directly upon the nature of the antigens that cause the 
homograft reaction. A state of tolerance, no matter how long 
it has prevailed, can be brought to an end simply by re- 
equipping the tolerant animal with normal, and therefore 
immunologically competent, lymph node cells. Consider an 
A-line mouse which has been made to accept a homograft of 
skin from a mouse of strain CBA, and let it be supposed that 
the homograft, bearing its characteristic coat of brown fur, 
has long been fully accepted by its host. The homograft can 
be destroyed, and the state of tolerance brought permanently 
to an end, simply by injecting the A-line mouse with normal 
lymph node cells taken from normal A-line donors. Until then, 
the antigens given forth by the CBA skin homograft were 
unable to elicit a reaction, because its recipients lymph node 
cells had been incapacitated by the CBA cells to which they 
were exposed in embryonic life. But when the A-line host has 
been refurnished with normal lymph node cells, then the 



antigens liberated by the graft have a chance to act upon 
lymphoid cells which are in normal working order; immunity is 
built up and the homograft is destroyed. The peculiar import- 
ance of this experiment is that it shows that the CBA skin 
homograft, though a perfectly normal tissue, was manufacturing 
antigens all the time — or rather, was manufacturing substances 
that would have been antigenic if only the host had been 
competent to recognize them for what they were. It is for this 
reason that we suppose that the emission of nuclear particles 
is a normal activity of living cells. The particles do nothing 
recognizable, and cannot even be shown to exist, unless the 
tissue from which they arise is transplanted to some other 
individual as a homograft. The nuclear particles then act as 
antigens and cause the tissue from which they originate to be 
destroyed. Yet it would be idle to suppose that this nuclear 
matter serves no function other than to make a nuisance of 
itself in the entirely artificial act of grafting; it merely so 
happens that grafting experiments of the design which I have 
just described provide at present the only method by which 
the nuclear matter discharged from cells can make its presence 
known. No more need be said here of the possible functions in 
normal life of the nuclear antigens which cause the homograft 
reaction, for I discussed them at length in Section 3. 

The problem of how tolerance comes about is still unsolved. 
For the present, the phenomenon of tolerance must be accepted 
as one of the raw data of immunology, and no theory of the 
mechanism of the immunological reaction wdll pass muster 
unless it can explain the phenomenon of tolerance as well. 
Haldane has suggested that an embryo has the power to 
metabolize — to break down and make use or dispose of — 
substances which the adult cannot metabolize; the adult makes 
antibodies against them instead. If that is so, then tolerance is 
the enforced retention of an embryonic modality of response; 
and it would be in keeping with Burnet and Fenner's theory if 



the transformation which leads to tolerance were closely akin 
to the ''training'' of bacteria to metabolize unaccustomed food- 
stuffs or to resist the action of inhibitory drugs. I think this 
interpretation is plausible, because to secure a state of toler- 
ance it is probably not sufficient merely to confront an embryo 
with antigens; the antigens must persist, and continue their 
educative action, well into the period in which, under normal 
circumstances, the young animal would have become immuno- 
logically mature. Fortunately, we need not commit ourselves 
to any particular theory of the mechanism of tolerance before 
examining its biological implications, which are various, 
though I can consider only one or two. 

The phenomenon of tolerance requires one to think anew 
about the nature of the relationship between a mammalian 
mother and her unborn young. Except in highly inbred animals, 
a foetus has a different genetic and therefore a different anti 
genie constitution from its mother. It is therefore an antigenic- 
ally foreign body, a kind of foreign graft. Why then does it not 
immunize the mother, with consequences disastrous to itself.'^ 

Haemolytic disease of the newborn is evidence that this does 
happen sometimes; that it happens very seldom is due to the 
extraordinarily efficient insulation of the mother from her 
unborn young. Under normal circumstances, no particles as 
large as cells, and probably no molecules as large as nucleo- 
proteins, could possibly cross the placental barrier from the 
foetus into the mother*'s blood. This arrangement provides 
against the danger that the foetus should immunize the mother, 
but this is not the only immunological danger, and a one-way 
control of traffic between foetus and mother is not enough. If 
particles a-s large as cells could pass from the mother to the 
foetus, then infective organisms or the antigens manufactured 
by them could also do so, and although maternal antibodies 
might keep infection in check, that would not prevent the 
antigens of micro-organisms from damaging, perhaps irrepar- 



ably, the future development of the immunological defences of 
the child. Nearly thirty years ago Traub showed that a virus 
disease of mice, lymphocytic choriomeningitis, can be trans- 
mitted from a mother to her unborn young; the young were 
accordingly unable to develop resistance to the virus in later 
life, and might transmit it to their young in turn. In this par- 
ticular strain of mice, virus and host had come to a live-and-let- 
live arrangement by which neither killed the other, and reflec- 
tion will show that, had this not been the case, the phenomenon 
could hardly have been discovered. But here is an example of a 
''hereditary'' infectious disease, running in a family, but trans- 
mitted from mother to young because each generation not 
merely infects its successor but abolishes its successor"'s power 
to rid itself of the disease. ''Genetic predisposition'' is therefore 
not the only possible explanation of a tendency of certain 
diseases to run in families. 

Under normal circumstances, the mere incorporation by a 
foetus of some of its mother"'s cells need not be expected to 
lead to evil consequences; it does not normally happen because, 
as I explained above, a frontier which lets through cells would 
let through undesirable immigrants as well. One can be con- 
fident that maternal cells are not admitted into the foetus, 
because if they were, the young should acquire a complete or 
partial tolerance of homografts transplanted from the mother, 
and this happens very rarely, if at all. Cancerous cells, however, 
are distinguished by their invasive properties, and there are 
half a dozen cases on record of the apparent transmission of 
malignant melanomatosis from a pregnant mother to her child. 
There can be little doubt that the melanoma cells actually 
crossed the placental frontier and established themselves in 
the foetus. An adult human being certainly, and even I think 
a newborn, will destroy homografts of malignant cells. It would 
not inevitably be disastrous for a foetus to allow, because it 
could not prevent, the growth of a foreign malignant tumour, 



provided only that it could rid itself of the tumour as soon as 
it became immunologically mature. Unfortunately, the effect 
of exposing the foetus to the malignant cells would be to 
prevent that very process of maturation, so that death early 
in post-natal life would be almost inevitable. 

When therefore we think of the immunological relationship 
between the mother and the foetus, we must read the relation- 
ship both ways round: the foetus must not be allowed to 
immunize the mother, and the mother must not be allowed to 
weaken the immunological defences of the child. It is for this 
reason, and for no other one sufficient reason, that the blood 
systems of the mother and the foetus must be strictly separate 
all the time, in every place, and at every level down to the finest 
capillary vessel. 

Beyond this, the concept of immunological tolerance has 
implications which are deeply philosophical, in the worst sense 
of the word, for it bears directly upon the problem of the 
recognition and awareness of The Self. Why do not the cells 
which undertake immunological responses react against con- 
stituents of the body in which they themselves are housed? 
W^hy are not ''auto-antibodies'' regularly formed? Alas for 
Naturphilosophie^ the problem is soluble and can be clearly put. 
The question of manufacturing auto-antibodies (or otherwise 
reacting) against antigens of the type that cause transplanta- 
tion immunity cannot arise in practice, because, as I have 
already explained, these antigens are uniformly represented in 
all the tissues of a single individual, not excepting his antibody- 
forming cells. Being part of the fabric of his o^vti antibody- 
forming cells, an individuaPs own ""transplantation antigens'* 
cannot be reacted upon as if they were foreign. But that does 
not explain why, for example, muscle protein, or any proteins 
distinctive of skin or nerve, should not appear foreign to 
an individuaPs own antibody-forming cells. The answer, we 
believe, is that his antibody-forming cells develop in the con- 



stant presence of, grow up with, these very substances, and so, 
in the technical sense I have just explained, become tolerant of 
what might otherwise have been their antigenic action. This 
explanation sounds too facile to be true, but fortunately there 
are certain exceptions which seem to prove the rule. Antibody- 
forming cells obviously cannot become tolerant of any bodily 
constituents which are formed, or do not become mature, until 
the antibody-forming cells themselves have formed and become 
mature; for example, they could not become tolerant of milk 
protein or chemically distinctive ingredients of spermatozoa. 
Nor could antibody-forming cells become tolerant of bodily 
constituents which, however early they develop, are physio- 
logically shut off from the remainder of the body, e.g. by lacking 
a blood supply or lymphatic drainage; they should not therefore 
become tolerant of the potentially antigenic action of the 
characteristic proteins of the lens. If this interpretation is 
correct, then substances like milk protein and lens protein and 
spermatozoa should be capable of forming auto-antibodies, 
though needless to say they never get a chance to do so in 
ordinary life. So they are; appropriately administered, all can 
form auto-antibodies in the body of which they themselves are 
part. The phenomenon of tolerance is therefore of fundamental 
importance in the mechanism by which the body learns to 
discriminate between what is proper to itself and what is 
foreign, and it is only under artificial or otherwise abnormal 
circumstances that the mechanism of recognition goes wrong. 


In this article I have shown how skin grafting can be used 
for the detection and assay of individuality, whether in gold- 
fish, mice or men. Although the inborn differences between 
human beings are combinational in origin and inner structure 
(they are not to be thought of as differences of either 'degree** 



or 'kind**), yet the combinants are so numerous, and so generous 
are the ways in which they may be combined, that every human 
being is genetically unique; the texture of human diversity is 
almost infinitely close woven. But what is the '"meaning"' of this 
diversity, i.e. what intelligible function does it fulfil? That is 
not a question one can very well ask of human beings, because 
the answer would be too complicated and too hedged around 
with qualifying clauses; but the gist of the answer, as it relates 
to lower organisms, is this. Inborn diversity makes for versatil- 
ity in evolution. Every living species must provide not only 
for the present but also for what may happen to it in the 
future; only those lineages survive to the present day which, 
in the past, were versatile enough to come to terms with their 
environment. All organisms must have a genetical system, as \ 
they must also have immunological and nervous systems, which 
can cope efiiciently with what has not yet been experienced — 
with what, if they were sentient, we should call the unforeseen. 
Bacteria and other micro-organisms, for example, must have a 
genetical system which will protect them as effectively from 
antibiotics which have vet to be discovered as from those which 
they have coped with hitherto. Only inborn diversity, and a ^ 
genetical system which keeps that diversity permanently in 
being, can make this possible. It is a mere truism that if inborn 
diversity and genetic individuality were to be extinguished, as 
in some animals they can be, by inbreeding, then selection 
would have nothing to act on, and the species would be left 
without evolutionary resource. Curiously enough, this would 
probably be less harmful to human beings than to any other 
animal, for men have devices for avoiding the rigours of selec- 
tion, and can change the environment instead of letting the 
environment change them. So far from being one of his higher 
or nobler qualities, his individuality shows man nearer kin to 
mice and goldfish than to the angels; it is not his individuality 
but only his awareness of it that sets man apart. 



Abercrombie, M. 21,86 

acquired characters. See La- 

adaptation 15, 84-9, 91 

bacterial 79-82, 102-4 

agammaglobulinaemia 153 

Agar, W. E. 95 

age-distribution 37, 59-60 

ageing 46. See also Senescence 

Algire, G. H. 159 

Alice, W.C. 22 

allergy 123, 174 

Allfrey, V. G. 168 

Allison, A. C. 126 

anabiosis 29 

Andjus, R. 30 

antibiotics 79-82, 144, 153, 185 

antibodies. See Immunity re- 

antigens. See Immunity reactions 

antiserum and eye defects 92-3 

Avery, O. T. 163 

bacteriophage 163 
Bailey, W.T. 102 
Baldwin, J. M. 82 
Baly, W. 31 
Bamatter, F. 147 
Barnes, D. W. H. 172 
Baskett, A. C. 103 
Basques 127 
Beadle, O. 128 
Beale, G. H. 100 

Beet, E. A. 126 

beha\^our 134-8 

human 134 

rats 94-6 

Bell, J. QQ 
Berkeley, G. 78 
Bernard, C. 17 
Bernheimer, A. W. 100 
Bidder, G. P. 25, 31-2, 57 
Billingham, R. E. 29-30, 84, m, 

130-1, 132, 161, 176-7 
blood groups 89, 123-7, 148, 

blood vessel grafts 169 
bone 87, 110, 129 
Borodin, N. A. 30 
Bowley, C. C. 151 
brain, mechanical 75, 140 
Brambell, F. W. R. 93 
Brent, L. 161, 177 
Briggs, R. 29 
Brown, G. W. 59 
Buffon, G. L. L. 17, 24 
Burnet, F. M. 175-6, 180 
burns 143-5, 173 
Burrows, M. T. 26 
Burt, W. H. SS 
Bushnell, L. D. 92 

carp 29 
Carrel, A. 26 
Carr-Saunders, A. M. 92 
Champy, C. 121 



Child, C. M. 29 
chimera 150-3, 173, 176-7 
Chitty, D. 33, 56 
chorea, Huntington's 66-7 
Clark, W. E. le G. 88, 114 
Comfort, A. 17, 22, 28 
cornea 86, 170-1 
corns 85-7 
cortisone 172-3 
Cowdry, E. V. 27 
Crew, F. A. E. 95 
Crick, F. H. C. 164 
cytoplasmic inheritance 100-5 

Darwin, C. 11, 84 

Darwinism 11-15,79,82,89-90 

fallacies about 11-15 

and Lamarckism 79-107 

death. See Senescence 

'natural' 18, 55-7 

de Beer, G. R. 12, 88 
Deevey, E. S. 22 
Delbruck, M. 102 
Descartes, R. 76 
diagnosis, clinical 74 
differentiation 24, 104, 121 
diffusion 113 

disk^, intervertebral 115,128-30 
domestication 33-4, 56, 63, 115 
Donald, H. P. 176 
Dowdeswell, W. H. 22 
Drew, J. S. 95 
Drummond, F. H. 95 
Dublin, L. I. 21, 42 
Dunsford, I. 151 


Eichwald, E. J 
Elton, C. 22 
embryo. See foetus 
Emerson, A. E. 22 

endosomatic instruments, evolu- 
tion 139-42 

Ephrussi, B. 100 

epidermis. See Skin 

eugenics 40 

evolution (see also Darwinism, 
Lamarckism) 11-6, 34-5, 1 22 3 

endosomatic, exosomatic 


experiment, defined 77-8 

Fenner, F. 175-6, 180 

Ferguson, K. A. 149 

fertility 14, 37-9, 54, 61-3, 153 

Finlay, G. F. 92 

fish, senescence in 31-2, 57 

sizes of 111-2, 117 

Fisher, R. A. 12, 14, 22, 37, 39, 

flexure lines 87, 89 
Flood, M. M. 59 
Flourens, M. J. P. 24 
Flower, S. S. 23, 32 
foetus 92-4, 123-5, 149, 181-3 
force of mortality. See Life Table 
Ford, C. E. 172 
Ford, E. B. 12, 22, 96 
form (see also Transformation) 

109, 117-21 
Fowler, E. H. 100 
Franceschetti, A. 147 
freemartinism 153 
freezing 29, 145-6 
Freund, J. 93 

Garstang, W. 12 

Geill,T. 26 

genes 28, 38-9, 63-70, 92, 101-2, 

126, 163-5 
Good, R. A. 153 



Gorer, P. A. 63, 158, 160 
graft hybrids. See Tolerance 
grafting 26, 41-2, 84-5, 89, 95, 

Griffith, F. 163, 166 
growth 108-21 

differential 54, 65, 114-21 

intussusceptive 131-2 

laws of 114 

rate 20-1, 47, 108-21 

Griineberg, H. 63 
Gunsen, M. M. 95 
Guyer, M. F. 92-3, 95 

Haeckel, E. 13 

haemolytic disease 123-7, 152, 

Haldane, J. B. S. 12, 28, 44, 67, 

92, 95, 113, 127, 180 
Hamerton, J. L. 172 
Harrison, J. A. 100 
Harrison, J. W. H. 96 
Harrison, R. G. 26 
Hasek, M. 151 
Helmholtz, H. L. F. von 122 
Hemmings, W. A. 93 
Henderson, M. 93 
heterografts 149, 155 
Hewer, H. R. 97 
Hildemann, W. H. 146 
Hinshelwood, C. N. 82, 102-3 
Hinton, M. A. C. 33 
Hogben, L. 21 
homografts. See Grafts 
homology 1 1 
Hughes, A. W. McK. 96 
Hunter, J. 29 
Huntington, J. 66-7 
Huntington's chorea 66-7 
Hurst, A. 25 

Hutchison, A. M. 151 
Huxley, J. S. 11, 65, 92, 118 
Huxley, T. H. 11 
Hyde, R. R. 93 
hypothermia 30 
hypothesis 72-4, 76-7 

Ibsen, H. L. 92 

immortality 26-30 

immunity reactions 42, 92-6, 99, 

123-4, 152-85 
induction, logical 73 
instinct 136-8 

intervertebral disks 115,128-30 
Irwin, M. R. 92 
isotopes 108 

Jackson, C. H. N. 22 
Jennings, H. S. 27 
Jensen, C. O. 26 
Johnson, M. L. 21 
joints 87-9, 128 
Jund, L. 29 

Kermack, K. A. 46 
Kilkenny, B. C. 103 
Kimball, R. F. 100, 102 
Klein, D. 147 
Kneale, W. 76 
Koga, Y. 23 
Korenchevsky, V. 17, 23 
Krohn, P. L. 41, 174 
Kukenthal, W. 88 

Lack, D. 14, 22, 56 
Lamarck, J. B. 79 
Lamarckism 13-4, 79-107 

defined 83, 91 

in miicro-organisms 99-105 

Landsteiner, K. 123 



Lankester, E. R. 11, 32-3, 57 

Lederberg, J. 102 

Lemche, H. 96 

Leslie, P. H. 21 

Levine, P. 123, 127 

life, expectation of 32-4, 42-5, 

49, 67 
life table 21-2,33-7,48-51, 58- 

Lindsley, D. L. 172 
Little, C. C. 160 
Lloyd Morgan, C. 82 
Loeb, L. 26, 95 
Lorenz, K. 125 
Lotka, A. J. 12, 21-2, 42, 139 
Loutit, J. F. 172 
Lovelock, J. E. 30 
lymph nodes 157, 161, 170, 179- 


MacBride, E. W. 25 

McCay, CM. 28 

McDougall, W. 94-8 

Mclndoe, A. 148 

malaria 126 

mammals, pregnancy 181-3 

senescence in 24-5, 32-5, 


sizes of 110-2 

Matthews, L. H. 112 
Medawar, P. B. 23-4, 86, 88, 

114, 161 
melanism 96-7 
melanocytes 86 
memory 54 
Mendel, G. 27 
Metalnikov, S. 19 
metatheory 74 
Metchnikoff, E. 17, 24, 25 
method, scientific 71-8 

Minot, C. 20-1, 23-4, 31, 47-8, 

Mirsky, A. E. 168 
Mitchell, P. C. 24 
Mitchison, N. A. 124, 157 
Morant, G. M. 23, 115 
mortality, force of 21-2, 33, 

36-9, 49-51 
Muller, H. J. 91 
MUller, J. 31 

Needham, J. 29 

Neel, J. V. 126 

nerve fibres 167-8 

Norton, H. T. J. 12 

Notestein, F. W. 42 

Nouy, L. du 26 

nucleic acid, nucleoprotein 162-6 

Odell, T. T. 172 
Olsen, E. 26 
Orla-Jensen, S. 26 
Owen, R. D. 150, 176 

paedomorphosis 12-13 
Palmer, J. F. 29 
Paramecium 100-4 
Park, O. 22 
Park, T. 22 
Parkes, A. S. 30 
Pauling, L. 126 
Pearl, R. 28, 30 
Pearson, K. 25, 35 
penicillin. See Antibiotics 
Penrose, L. S. 44, 67 
Pierson, B. F. 28 
placenta. See Foetus 
polymorphism 89, 125-7 
Popper, K. 16 
Poulton, E. B. 17 



pregnancy. See foetus 
Prehn, R. T. 159 
Protozoa, immortality of 27 
inheritance in 105 

Race, R. R. 151 

Ranson, R. M. 21 

rats, growth and longevity 29 

inheritance in 94-6 

recapitulation, law of 13 
reproductive value 37, 39, 60-1, 

Rhine, J. B. 94-5 
Rowlands, H. A. 75 
Russell, B. A. W. 74 
Russell, P. S. 132 

Sanger, R. 151 
Schinkel, P. G. 149 
Schmidt, K. P. 22 
Schmorl, G. 129 
Schonland, S. 18 
Schrodinger, E. 14 
scientific method 71-8 
selection (see also Darwinism.) 

79-82, 94-6, 102-4, 169 
senescence 17-43, 44-70, 129 

definition of 22, 55 

evolution of 17-43, 44-70 

measurement 47-55 

Shipley, A. E. 18 

sickle cell disease, trait 125-6 

Silmser, C. R. 148 

Simonsen, M. 177 

Simpson, G. G. 34 

size 31-2, 110-7 

skin, ageing in 52-3 

grafting 143-185 

healing of 130-3, 143-6 

varieties of 84-7 

Sladden, D. 97 

Smith, A. U. 30 

Smith, F. A. 92, 93, 95 

Smith, J. M. 112 

Snell, G. D. 160 

Sonneborn, T. M. 28, 100, 105 

Spencer, H. 79, 112, 113 

Spencer's Law 112-3 

Spiegelman, M. 21, 42 

spinal column 127-30 

squatting facets 88 

stick insects, inheritance in 

Strangeways, T. 27 
Sturtevant, A. H. 92-3 

Tanner, J. M. 115 
Tatum, E. L. 102 
Tausche, F. G. 172 
Thompson, D'A. W. 11, 118, 

Thompson, J. S. 151 
Thompson, W. S. 42 
Thomsen, M. 96 
Thorpe, W. H. 98 
Tiegs, O. W. 95 
time, biological 29 
Tinbergen, N. 135 
tolerance, immunological 152, 

160, 175-184 
tradition 134-142 
training of bacteria 79-82,102-4 

of rats 94-6 

transformations, bacterial 163, 


Thompsonian 120-1 

transplantation. See Grafting 

Traub, E. 182 

Trotter, W. 140 

twins 147-52, 155, 176-7 





ultrasonic irradiation 161-2 

Varco, R. L. 153 
vertebral column 127-30 

Waddington, C. H. 12, 84, 86, 

Wallace, A. R. 18, 79 
Watson, J. D. 164 
Weaver, J. M. 159 
Webster, J. P. 29 
Weismann, A. 18-20, 30, 34, 

36-7, 40, 57-8, 70 
Whelpton, P. K. 42 
Whitear, M. 54 

Whitehead, A. N. 137-8 

Wiener, A. S. 123 

Woglom, W. H. 26 

Wong, H. 127 

Woodger, J. H. 79 

Wood-Jones, F. 88 

wound healing 47, 130-3, 143-6 

Wright, S. 12 

wrinkles 52-3 

X-rays 172-3 

Young, J. Z. 55 

Zuckerman, S. 121